1.2.15 (2020-10-10)

• check if template/ direcotry exists, because this is auto-generated directory (#2537)
  • install within roseus_FOUND
  • check if template/ direcotry exists, because this is auto-generated directory
• Contributors: Kei Okada

1.2.14 (2020-10-09)

• remove packages=['jsk_perceptoin'] (#2536)
• fix

```

System Message: WARNING/2 (<string>, line 32); backlink

Inline literal start-string without end-string.

System Message: WARNING/2 (<string>, line 32); backlink

Inline interpreted text or phrase reference start-string without end-string.

• /usr/bin/env PYTHONPATH=/opt/ros/melodic/lib/python2.7/dist-packages:/tmp/jsk_recognition-release/obj-x86_64-linux-gnu/lib/python2.7/dist-packages:/opt/ros/melodic/lib/python2.7/dist-packages:/home/k-okada/pynaoqi/pynaoqi-python2.7-2.5.5.5-linux64/lib/python2.7/site-packages CATKIN_BINARY_DIR=/tmp/jsk_recognition-release/obj-x86_64-linux-gnu /usr/bin/python2 /tmp/jsk_recognition-release/setup.py egg_info --egg-base /tmp/jsk_recognition-release/obj-x86_64-linux-gnu build --build-base /tmp/jsk_recognition-release/obj-x86_64-linux-gnu install --root=/tmp/jsk_recognition-release/debian/ros-melodic-jsk-perception --install-layout=deb --prefix=/opt/ros/melodic --install-scripts=/opt/ros/melodic/bin

System Message: WARNING/2 (<string>, line 21)

Bullet list ends without a blank line; unexpected unindent.

running egg_info creating /tmp/jsk_recognition-release/obj-x86_64-linux-gnu/jsk_perception.egg-info writing /tmp/jsk_recognition-release/obj-x86_64-linux-gnu/jsk_perception.egg-info/PKG-INFO writing top-level names to /tmp/jsk_recognition-release/obj-x86_64-linux-gnu/jsk_perception.egg-info/top_level.txt writing dependency_links to /tmp/jsk_recognition-release/obj-x86_64-linux-gnu/jsk_perception.egg-info/dependency_links.txt writing manifest file '/tmp/jsk_recognition-release/obj-x86_64-linux-gnu/jsk_perception.egg-info/SOURCES.txt' error: package directory 'jsk_perception' does not exist CMake Error at catkin_generated/safe_execute_install.cmake:4 (message): execute_process(/tmp/jsk_recognition-release/obj-x86_64-linux-gnu/catkin_generated/python_distutils_install.sh) returned error code Call Stack (most recent call first): cmake_install.cmake:41 (include) Makefile:97: recipe for target 'install' failed

```

• Contributors: Kei Okada

1.2.13 (2020-10-08)

• fix logic to check chainer version (#2534)
  • add test to check #2533 regression
• Contributors: Kei Okada

1.2.12 (2020-10-03)

• check if chainer is found before check version (#2533)
  • fixes http://build.ros.org/job/Nbin_uF64__jsk_perception__ubuntu_focal_amd64__binary/1/console and http://build.ros.org/job/Mbin_uB64__jsk_perception__ubuntu_bionic_amd64__binary/91/console
• Contributors: Kei Okada

1.2.11 (2020-10-01)

• Add FCN8sDepthPredictionConcatFirst model to fcn_depth_prediction.py (#2481)

  • Update output file names
  • Read dataset directory from argument
  • Fix directory name of dataset extracted from tar ball
  • Flatten images for network input
  • Remove wrong transform of dataset from train_fcn_depth_prediction.py
  • Add training script of FCNDepthPredictionConcatFirst model
  • Move FCN8sDepthPrediction chainer models to jsk_recognition_utils
  • Add install script of mirror dataset
  • Fix typo of model path
  • Use cv2 version of colormap JET
  • Add trained model of FCN8sDepthPredictionConcatFirst
  • Add FCN8sDepthPredictionConcatFirst model to fcn_depth_prediction.py

• refactor sample launches in jsk_perception (#2376)

• Add nose mask publisher (#2347)

• [jsk_perception/people_pose_estimation_2d.py][jsk_perception/people_mask_publisher.py] Fix edge case bug (#2465)

• Publish ClusterPointIndices in ssd_object_detector.py (#2467)

  • add predict profilling message above cluster indices computation

• fix travis - skip noetic test into two jobs, using BUILD_PKGS - skip catkin_python_setup for indigo (#2522)

• Fix for noetic / 20.04 (#2507)

  • jsk_perception/scripts: respect ROS_PYTHON_VERSION
  • support for opencv4 : jsk_perception
  • remove signals from find_package(Boost)
  • jsk_perception depends on roseus, but it sometimes hard to keep dependency
  • fix for python3, use 2to3 -f print, 2to3 -f except
  • upgrade package.xml to format=3, migrate to noetic with ROS_PYTHON_VERSION=2/3, use multiple ROS distro strategy http://wiki.ros.org/noetic/Migration

• more fix for #2500 (#2502)

  • fix print '' -> print('')

• fix print syntax in train_ssd.py (#2500)

  • fix print '' -> print('')

• [jsk_perception] support image with alpha in image_publisher (#2479)

  • fix image_publisher for loading grayscale image
  • use cv2 default type
  • add test for image with alpha channel
  • add sample for alpha image
  • fix for depth image
  • support image with alpha in image_publisher

• [jsk_perception] add program for training ssd with box annotation (#2483)

• show what should we do, if we have error on 'import chainer' (#2491)

  • use --clock for sample_image_cluster_indices_decomposer.launch, add --clock to sample_bounding_box_to_rect.launch does not work...
  • print how to intall cupy

  System Message: WARNING/2 (<string>, line 110)

  Bullet list ends without a blank line; unexpected unindent.

  if you do not have cupy, it raises error ` [INFO] [1588763738.839739]: Read the image file: /home/k-okada/ws_recognition/src/jsk_recognition/jsk_perception/sample/object_detection_example_2.jpg [INFO] [1588763739.625133]: Loaded 43 labels Traceback (most recent call last): File "/home/k-okada/ws_recognition/src/jsk_recognition/jsk_perception/node_scripts/ssd_object_detector.py", line 207, in <module> ssd = SSDObjectDetector() File "/opt/ros/melodic/lib/python2.7/dist-packages/jsk_topic_tools/transport.py", line 26, in __call\_\_ obj = type.__call_\_(cls, *args, **kwargs) File "/home/k-okada/ws_recognition/src/jsk_recognition/jsk_perception/node_scripts/ssd_object_detector.py", line 71, in __init\_\_ chainer.cuda.get_device_from_id(self.gpu).use() File "/usr/local/lib/python2.7/dist-packages/chainer/backends/cuda.py", line 275, in get_device_from_id check_cuda_available() File "/usr/local/lib/python2.7/dist-packages/chainer/backends/cuda.py", line 138, in check_cuda_available raise RuntimeError(msg) RuntimeError: CUDA environment is not correctly set up (see https://github.com/chainer/chainer#installation).No module named cupy `` * show what should we do, if we have error on 'import chainer' ` Traceback (most recent call last): File "/home/k-okada/ws_recognition/src/jsk_recognition/jsk_perception/node_scripts/ssd_object_detector.py", line 26, in <module> import chainer File "/usr/local/lib/python2.7/dist-packages/chainer/__init__.py", line 10, in <module> from chainer import backends # NOQA File "/usr/local/lib/python2.7/dist-packages/chainer/backends/__init__.py", line 1, in <module> from chainer.backends import cuda # NOQA File "/usr/local/lib/python2.7/dist-packages/chainer/backends/cuda.py", line 77 def shape(self) -> types.Shape: ^ SyntaxError: invalid syntax ``` c.f. https://github.com/jsk-ros-pkg/jsk_recognition/pull/2485

  System Message: WARNING/2 (<string>, line 110); backlink

  Inline literal start-string without end-string.

• add more arg INPUT_IMAGE (#2492)

  • arg name='INPUT_IMAGE' need to use default, instead of value, so that we can cheange the input name as ros args. 'value' is constant value and 'default' is default value, see http://wiki.ros.org/roslaunch/XML/arg

• jsk_perception/train_ssd.py fix error when out_dir is set (#2493)

• set chainer version less than 7.0.0 (#2485)

  • split test_bing to test_bing_output and test_bing_objectness
  • add time-limit to jsk_pcl_ros/test/test_linemod_trainer.test, jsk_perception/test/bing.test
  • jsk_perception/package.xml: node_scripts/pointit.py imports tf2_geometry_msgs
  • set time-limit=25 for timeout:30 tests
  • relax test conditions
  • set chainer version less than 7.0.0
  • jsjk_perception/train_ssd.py fix error when out_dir is set

• Fix test for consensus_tracking (#2475 from YutoUchimi/fix_consensus_tracking

• Parameterize frames, transformation and interpolation in virtual_camera_mono (#2470)

  • Change logger level of TransformException to WARN
  • Add test for virtual_camera_mono
  • Add sample for virtual_camera_mono
  • Parameterize virtual_camera_mono

• Convert audio data to spectrogram (#2478)

  • add unit to axis
  • remove unused files
  • add node to visualize spectrum
  • fix size of spectrogram
  • fix typo in launch
  • divide program into audio_to_spectrum and spectrum_to_spectrogram
  • fix comment
  • add test
  • use rosbag with /audio of 300Hz
  • use timer callback to publish spectrogram constantly
  • update comments and name of parameter
  • add sample program to convert audio message to spectrogram

• Add train script and sample for SSD (#2471)

  • [jsk_perception] add program for training ssd with box annotation
  • use cv2 for cv_resize_backend
  • add classnames for ssd
  • add trained model in install_trained_data.py

• Add queue_size and slop param to TileImages (#2453)

• Fix label_id division by 256 -> 255 (#2455)

  • Fix label_id division by 256 -> 255 Since len(colormap) is 255, % 256 is wrong since it can return 255 which raises IndexError.

• fix generate_readme.py and update readme (#2442)

• Publish human skelton msgs in OpenPose node (#2437)

  • add lines considering shoulder when predicting face region
  • add LIMB_PART param
  • enable to create nose mask image
  • [jsk_perception/node_scripts/people_pose_estimation_2d.py] fix edge case
  • [jsk_perception/node_scripts/people_mask_publisher.py] fix edge case

• Fix tile_image.py for Python3 (#2452)

• Fix label_image_decomposer.py for Python3 (#2454)

• Update to slic d77d6e8 (#2450)

• mask_rcnn_instance_segmentation: support loading yaml from file (#2413)

• pointit: add option '~use_arm' to select arm for pointing (#2415)

• Add sample, test and doc (#2440)

  • Fix condition of fatal message
  • Keep backward compatibility for ~dist_threshold
  • Add test for kalman-filtered-objectdetection-marker.l
  • Add sample for kalman-filtered-objectdetection-marker.l
  • Change permission of kalman-filtered-objectdetection-marker.l: 644->755
  • Update sample for RobotToMaskImage
  • Add sample for CollisionDetector
  • Merge branch 'master' into kinfu-fix
  • updae people_pose_estimation_2d.test
  • add visualization link in commentout
  • Add test for RobotToMaskImage
  • Add minimal sample for RobotToMaskImage, which is only for testing
  • Add test for SlidingWindowObjectDetector
  • Add sample for SlidingWindowObjectDetector
  • Support overriding parameters in manifest file
  • Add sample for sliding_window_object_detector_trainer_node
  • Add params for fg/bg training dataset image topics and output manifest file
  • Add test for ColorHistogramLabelMatch
  • Add sample for ColorHistogramLabelMatch
  • Add test SingleChannelHistogram
  • Add sample for SingleChannelHistogram
  • Explicitly depend of topic_tools because sample_polygon_array_color_histogram.launch uses this
  • Add test for PolygonArrayColorLikelihood
  • Add sample for PolygonArrayColorLikelihood
  • Suppress very long log of downloading pretrained weight in sample_deep_sort_tracker.launch
  • Add test for PolygonArrayColorHistogram
  • Add sample for PolygonArrayColorHistogram
  • Support selecting histogram index by rosparam in unwrap_histogram_with_range_array.py
  • Build SnakeSegmentation only when OpenCV<3
  • Add test for UnapplyMaskImage
  • Add sample for UnapplyMaskImage
  • Add test for TabletopColorDifferenceLikelihood
  • Add sample for TabletopColorDifferenceLikelihood
  • Add test for SnakeSegmentation
  • Add sample for SnakeSegmentation
  • Add test for Skeletonization
  • Add sample for Skeletonization
  • Add test for SaliencyMapGenerator
  • Add sample for SaliencyMapGenerator
  • Add test for ROIToRect
  • Add sample for ROIToRect
  • Fix output polygon vertices for ROIToRect
  • Add test for ROIToMaskImage
  • Add sample for ROIToMaskImage
  • Add test for RectToROI
  • Add sample for RectToROI
  • Add test for RectToMaskImage
  • Fix ROSTimeMoveBackward before publishing output in sample_rect_to_mask_image.launch
  • Fix point index for bottom right point of rectangle in rect_to_mask_image.cpp
  • Add test for ProjectImagePoint
  • Add sample for ProjectImagePoint
  • Add test for PolygonToMaskImage
  • Add sample for PolygonToMaskImage
  • Add test for PolygonArrayToLabelImage
  • Add sample for PolygonArrayToLabelImage
  • Add test for MaskImageToROI
  • Add sample for MaskImageToROI
  • Add test for GrabCut
  • Add sample for GrabCut
  • Disable fast_rcnn.test
  • Add test for FisheyeToPanorama
  • Add sample for FisheyeToPanorama
  • Add test for GaussianBlur
  • Add sample for GaussianBlur
  • Add test for YCCDecomposer
  • Add sample for YCCDecomposer
  • Add test for LabDecomposer
  • Add sample for LabDecomposer
  • Add test for RGBDecomposer
  • Add sample for RGBDecomposer
  • Add test for HSVDecomposer
  • Add sample for HSVDecomposer
  • Add test for morphological operators
  • Add sample for morphlogical operators such as ErodeMaskImage, Opening, MorphlogicalGradient, TopHat
  • Add test for pointit.py
  • Add sample for pointit.py
  • Remove unused import in pointit.py
  • Remove unused computation in get_marker func in pointit.py
  • Fix tf2 listener
  • Fix return value in find_pose func in pointit.py
  • Add test for unwrap_histogram_with_range_array.py
  • Add sample for unwrap_histogram_with_range_array.py
  • Add test for solidity_rag_merge.py
  • Add sample for solidity_rag_merge.py
  • Support networkX>=2 and scikit-image>=0.13 in solidity_rag_merge.py
  • Add test for non_maximum_suppression.py
  • Add sample for non_maximum_suppression.py
  • Add ROS topic API for non_maximum_suppression.py
  • pointit: add option '~use_arm' to select arm for pointing
  • mask_rcnn_instance_segmentation: support loading yaml from file
  • add jsk_perception/SubtractMaskImage
  • fix typo in sample_face_pose_estimation.launch
  • GPU -> gpu in face_pose_estimation.launch
  • use args in sample launch: GPU -> gpu
  • remove test_mode from sample_face_pose_estimation.launch
  • remove test_mode in sample_ssd_object_detector.launch
  • Do not use deprecated param in sample_pointit.launch
  • Fix use of deprecated param ~dist_threshold

• fixes scope bug on point_pose_extraction (#2414)

• [jsk_perception] Add trained maskrcnn model for 73b2 kitchen (#2423)

  • update kitchen pretrained model (#9)
  • [jsk_perception] Add trained maskrcnn model for 73b2 kitchen
  • add sample launch file using 73b2 kitchen model
  • update kitchen pretrained model
  • add sample launch for kitchen dataset

• update to use jsk_travis 0.5.0 (#2439) * skip deep_sort_tracker.test on indigo https://travis-ci.org/jsk-ros-pkg/jsk_recognition/jobs/549216064#L8697-L8733 downloading SSD data(ssd300_voc0712_converted_2017_06_06.npz) failes with ` IOError: [Errno socket error] [Errno 1] _ssl.c:510: error:14077410:SSL routines:SSL23_GET_SERVER_HELLO:sslv3 alert handshake failure' ` do we need to update Python to 2.7.9? for indidgo ???? https://stackoverflow.com/questions/54413685/insecureplatform-warning

  • Do not mix tab and space for indentation
  • Add test for mask_rcnn_instance_segmentaion.py, but comment out testing because GPU required
  • Add test for image_time_diff.py
  • Add sample for image_time_diff.py
  • Avoid crashing when ROS time moved backward in image_time_diff.py
  • Fix AttributeError in image_time_diff.py
  • Add test for fcn_depth_prediction, but do not run because unstable
  • Add test for fast_rcnn.py
  • Add test for binpack_rect_array.py
  • Add sample for binpack_rect_array.py
  • Add test for apply_context_to_label_probability
  • Add gpu arg to sample_apply_context_to_label_probability.launch
  • fix typo: skelton -> skeleton
  • publish skelton in people_pose_estimation_2d

• Add Mask R-CNN model trained with COCO dataset (~80 classes) (already included VOC model only detects ~20 classes) (#2427)

• MaskImageToPointIndices: support multi channel mask image (#2409)

  • fix mask rcnn 73b2 model classname typo (#8)

• point_pose_extractor: fix bug on scope

• point_pose_extractor: fill reliability

• Add sample for MaskImageToPointIndices

• add jsk_perception/SubtractMaskImage (#2411)

  • Fix typo of main node name

• Re-enable bing.test (#2418)

  • Fix target name of bing for testing

• Contributors: Fuki Furuta, Kei Okada, Kentaro Wada, Naoya Yamaguchi, Shingo Kitagawa, Yoshiki Obinata, Yuki Furuta, Yuto Uchimi, Iory Yanokura, Hideaki Ito, Taichi Higashide

1.2.10 (2019-03-27)

• Fix error on setting device number other than 0 on multiple gpu env. (#2412)
  • face_pose_estimation: support multi gpu env mask_rcnn_instance_segmentation.py: support multi gpu env people_pose_estimation_2d.py: support multi gpu env ssd_object_detector.py: support multi gpu env
• Re-enable draw_classification_result.test (#2401)
  • Re-enable draw_classification_result.test
  • Increase slop for bof_histogram_extractor
• Re-enable color_histogram.test( #2400)
• Contributors: Yuki Furuta, Yuto Uchimi

1.2.9 (2019-02-23)

1.2.8 (2019-02-22)

1.2.7 (2019-02-14)

• [jsk_perception/ssd_object_detector.py] Add header for publishing result image (#2367 )
• [jsk_perception] Add deep_sort_tracker_node.py (#2351 )
  • [jsk_perception/deep_sort_net.py] Fixed deep_sort_net import
  • [jsk_perception/test/deep_sort_tracker.test] Disable gpu in test
  • [jsk_perception/sample_deep_sort_tracker.launch] Refactor
  • [jsk_perception/deep_sort_tracker_node.py] Modified import file not to depend on tensorflow
  • Revert "[jsk_perception/deep_sort_tracker] Add dependencies of tensorflow" This reverts commit 7dac944cfc9292d81b8bdb90d89e8100eda2bf3a.
  • [jsk_perception/deep_sort_tracker] Add dependencies of tensorflow
  • [jsk_perception/deep_sort_tracker] Install git submodule directory to node_scripts/deep_sort/deep_sort
  • [jsk_perception/deep_sort_tracker_node.py] Add target_labels param to specify input labels/recst
  • [jsk_perception/deep_sort_tracker_node.py] Add test
  • [jsk_perception/deep_sort_tracker_node.py] Renamed publish image topic vis -> viz
  • [jsk_perception/deep_sort_tracker_node.py] Publish labelarray
  • [jsk_perception/sample/deep_sort_tracker] Add pretrained model load
  • [jsk_perception/deep_sort_tracker_noder.py] Add node
  • [jsk_perception/deep_sort_tracker_node.py] Add sample
  • [jsk_perception/install_trained_data.py] Add deepsort trained model
  • [jsk_perception] Add deep_sort by gitsubmodule
• [doc] [jsk_perception] Add documentation (#2385 )
  • Rewrite matchtemplate.py with cv2
  • Add test for matchtemplate.py
  • Enable random_forest_server.test only in indigo.
  • Add sample for matchtemplate.py
  • Fix conversion for latest cv_bridge: imgmsg <-> cv2 <-> cv
  • Add test for fisheye_ray.py
  • Add sample for fisheye_ray.py
  • Fix for undefined global variable in fisheye_ray.py
  • Add test for random_forest_server
  • Publish ~output/debug_image in random_forest_client_sample.py
  • Fix for executing RandomForestClassifier
  • Remove unused sklearn module which causes ImportError in sklearn>=0.20
  • Add ~slop param to bof_histogram_extractor
  • Show viewer if gui:=true in sample_background_subtraction
  • Remove unused remapping in sparse_image.test
  • Fix sparse_image_encoder/decoder sample
• [jsk_perception] Support fcn8s_atonce model in fcn_object_segmentation.py (#2375 )
  • Fix typo: fcn8s_atonce -> fcn8s_at_once
  • Support fcn8s_atonce model in fcn_object_segmentation.py
• [jsk_perception] fix load path for kalmanlib.l (#2377 )
• [doc] [jsk_perception] [jsk_recognition_utils] Add guide to image recognition with deep learning (#2365)
  • Add doc for image annotation
  • Add annotate_images_with_labelme to index
  • Add dataset class for semantic segmentation
  • Add install_learning_datasets script
  • Download datasets during catkin build
  • Add .gitignore in learning_datasets/
  • Add train_fcn script
  • Set default learning_rate to valid value
  • Enable plotting from remote host as well
  • Add doc for training FCN
  • Add doc for starting deep learning with image dataset
  • Add how to create dataset, where to store it in documentation
  • Dump param for fcn_object_segmentation.py
  • Add InstanceSegmentationDataset
  • Add train script for Mask-RCNN
  • Fix model_name and outputs in train_fcn.md
  • Add doc for training Mask-RCNN
• Contributors: Kei Okada, Yuki Furuta, Yuto Uchimi, Iori Yanokura

1.2.6 (2018-11-02)

• Add hand pose detection (#2324) * [jsk_perception/people_pose_estimation.py] Fixed for cpu inference * [jsk_perception/people_pose_estimation.py] Diable train and enable_backprop * [jsk_perception/people_pose_estimation_2d] Add hand width offset * pointit: add handle exception on tf2 * pointit: add min threshold * jsk_perception: add pointit * people_pose_estimation_2d: support hand detection

• [jsk_perception] Add human mesh recovery(estimate people 3d pose from 2d image) (#2332) * clean up jsk_perception/scripts/install_trained_data.py around if _chainer_available * [jsk_perception/human_mesh_recovery] Refactor * [jsk_perception/human_mesh_recovery] Add test * [jsk_perception/human_mesh_recovery] Add sample * [jsk_perception/human_mesh_recovery] Add install model file code * [jsk_perception/human_mesh_recovery] Add node

• [jsk_perception/openpose] Add resize image (#2300) * [jsk_perception/openpose] Fixed logic * [jsk_perception/openpose] Add warning * [jsk_perception/openpose] Add resize image

• [jsk_perception/ssd_object_detector] Add hand pretrained model (#2333)

• Fix install destination (#2345) * Install 'node_scripts', 'scripts', 'test' into SHARE_DESTINATION

• [jsk_perception/sample_mask_rcnn] Fixed typo. fps -> rate (#2353)

• [jsk_perception/mask_rcnn_instance_segmentation.py] Publish rects and class (#2350)

• [jsk_perception/point_pose_extractor.cpp] Correct grammer. 'could not found' -> 'could not find' (#2349)

• [jsk_perception/image_publisher.py] Add fov parameter for publishing valid camera info parameters (#2340) * [jsk_perception/image_publisher.py] Add warning when not specified fovx and fovy at the same time * [jsk_perception/sample_image_publisher.launch] Add fov parameter for kinectv2 * [jsk_perception/image_publisher.py] Add fov parameter for camera info

• [jsk_perception/sample_bof_object_recognition.launch] Fixed path of trained bof data(#2337) * [jsk_perception/install_trained_data.py] Add trained bof data for sklearn==0.20.0

• fix for jsk-ros-pkg/jsk_common/pull/1586 (#2311) * to avoid add_custom_target cannot create target install_sample_data because another target with the same name already exists errors

• Use diagnostic nodelet for EuclideanClustering and other nodelets (#2301) * jsk_pcl_ros: euclidean_clustering: use dianogistc nodelet

  System Message: ERROR/3 (<string>, line 510)

  Unexpected indentation.

  Use DiagnosticNodelet::updateDiagnostic preferrably

• support SSD512 for ssd_object_detector (#2305) * move ssd_train_dataset to scripts

• [jsk_perception/face_pose_estimation] Fixed orientation of face pose (#2304) * [jsk_perception/face] Modified rviz * [jsk_perception/face] Add debug image of face pose * [jsk_perception/face] Fixed orientation of publish pose * [jsk_perception/face] Fixed pretrained model loader

• Enable Openpose Node for chainer 4.0.0 (#2295) * [jsk_perception/scripts] Modified url * [jsk_perception/scripts] Modified format * [jsk_perception/scripts] Modified openpose's weight * [jsk_perception] Modified openpose

• [jsk_perception] install config dir (#2294)

• Update chainer_mask_rcnn to 0.3.0 (#2293

• Fix for AssertionError in fast_rcnn.py (#2281) * Ignore whether cuda is available or not in fast_rcnn.py * Allow ~gpu as rosparam in fast_rcnn * Fix for AssertionError in fast_rcnn.py

• Re-enable tests which use chainer inside them (#2280) * Re-enable all tests which use chainer * Re-enable tests which use chainer inside them

• Set required=true for samples to fast finish in testsMerge pull request (#2274)

• Refactor cmake of jsk_perception (#2275) * Apply Eigen -> Eigen3 migration (Eigen also works) http://wiki.ros.org/jade/Migration * Remove no need libsiftfast dependency

• fix travia and reduce dependency for jsk_pcl_ros (#2276) * skip test for #2272 * Set required=true for samples to fast finish in tests

  System Message: ERROR/3 (<string>, line 546)

  Unexpected indentation.

  Sometimes the test fails because of unexpected errors. In that case, it is better that the test quickly finish with errors.

  System Message: WARNING/2 (<string>, line 548)

  Block quote ends without a blank line; unexpected unindent.

  • skip more tests

• Contributors: Yuki Furuta, Kei Okada, Kentaro Wada, Riku Shigematsu, Shingo Kitagawa, Yuto Uchimi, Iori Yanokura

1.2.5 (2018-04-09)

• Add MaskRCNNInstanceSegmentation node (#2257) * MaskRCNN node publishes label imgs that have class and instance id * Add ~bg_label to label_image_decomposer which is not colorized * Add ~cval param to apply_mask_image * Add MaskRCNNInstanceSegmentation node

• Improve topic name visualization in tile_image.py (#2256) * Comment out draw_classification_result test * Improve visualization in tile_image.py

  System Message: ERROR/3 (<string>, line 563)

  Unexpected indentation.

  • Use FONT_HERSHEY_SIMPLEX.
  • Adjust font_scale according to the new font.

• [jsk_perception/draw_classification_result.py] use LINE_AA for opencv3 in kinetic (#2247) * enable draw_classification_result test * remove unused variables and imports * use LINE_AA for opencv3 in kinetic

• Add fcn_depth_prediction node (#2244) * [jsk_perception] Fix function name in fcn_depth_prediction.py * [jsk_perception] Add sample of fcn_depth_prediction * [jsk_perception] Add trained data for fcn_depth_prediction to install_trained_data * [jsk_perception] Add fcn_depth_prediction node

• [jsk_perception/fast_rcnn.py] fast_rcnn node to follow chainer-v2 version (#2249) * add MODEL arg for fast rcnn launch * check chainer version for volatile variable

• [jsk_perception/label_image_decomposer.py] check img.ndim for gray scale image (#2248) * check img.ndim for gray scale image

• Contributors: Yuki Furuta, Kei Okada, Kentaro Wada, Shingo Kitagawa, Yuto Uchimi

1.2.4 (2018-01-12)

• jsk_perception: install template dir (#2222)
• Contributors: Yuki Furuta

1.2.3 (2017-11-23)

• jsk_perception: add face_pose_estimation (#2207)

• jsk_perception: people_pose_estimation_2d.py: add option not to synchronize camera info

• jsk_perception: use 'find' in generated eusmodel launch file (#2215)

• add timestamp for diff_image (#2216)

• jsk_percetion: add ssd object detector (#2204 from furushchev/ssd)

• Drop hydro from CI on Travis (#2217) * Remove color_histogram test that won't work on Travis

• Capability of specifying shape for tiling images (#2208) * Refactor tile_image.py about self._shape * Validate ~shape param of tile_image.py

  System Message: ERROR/3 (<string>, line 604)

  Unexpected indentation.

  • modified: tile_image.py

  System Message: WARNING/2 (<string>, line 605)

  Block quote ends without a blank line; unexpected unindent.

  • Capability of specifying shape for tiling images

• Add ~alpha param to label_image_decomposer to tune the overlay (#2211) * Make ~alpha param of label_image_decomposer to dynparam * Add ~alpha param to label_image_decomposer to tune the overlay * Add option to visualize label image without sync by ~only_label option

• jsk_perception: people_pose_estimation_2d.py: unsynchronize camera info (#2206)

• Add node for visualization of (labeled) rectangle region on 2D image (#2205) * jsk_perception: draw_rects: disable resubscribing on hydro * jsk_perception: use jsk_recognition_msgs::Rect for rect instead of geometry_msgs::PolygonStamped * jsk_perception: add nodelet for drawing rects on image * jsk_perception: use classification result for FastRCNN

• Split test of fcn_object_segmentation to avoid MemoryError Because loading 2 FCN8s model is too heavy on PCs with small memories. (#2200)

• [jsk_perception, slic_super_pixels] add parameter, publish_debug_images (#2181)

• Regional feature based object recognition using ResNet (#2172) * Rename to regional_feature_based_object_recognition * Remove params pretrained_model and mean_file * Sort add_rostest * Add test for feature_based_object_recognition * Download files and make the sample work * Add ResNetFeature * Fix bug in feature_based_object_recognition * Add feature based object recognition node * Large color variation in draw_classification_result * Display image even though some topics have not come yet * Fix nan values in ProbabilityImageClassifier

• node_scripts/apply_context_to_label_probability: make sure candidates is list because it can be tuple, which cause error (#2185)

• Fix ignore_labels out of range for the input label/proba image (#2184) * Update sample of label/probability_image_classifier

• Fixes on probabilistic image classifier (#2177) * If no candidates, candidates_fixed should be ignored

• src/bounding_box_to_rect.cpp: Convert bounding box to mask (#2176) * Add sample_rect_to_mask_image.launch * support BoundingBox as input topic type as well as BoundingBoxArray

• jsk_perception: fix indent in creating people pose (#2179)

• Contributors: Yuki Furuta, Kei Okada, Kentaro Wada, Naoki Hiraoka, Shingo Kitagawa, Yohei Kakiuchi, Yuto Uchimi

1.2.2 (2017-07-23)

• add bg_label in apply_context_to_label_probability (#2175) * Remove no need ~use_topic flag * Refactor to handle fixed candidates in ApplyContextToLabelProbability * add bg_label in apply_context_to_label_probability
• fix bug in label_image_classifier (#2174) * Update label_image_classifier.py * fix bug in label_image_classifier
• Contributors: Kentaro Wada, Shingo Kitagawa

1.2.1 (2017-07-15)

• If chainer is not installed, use v2 (#2167) * chainer can not install in ros build firm
• Contributors: Kei Okada

1.2.0 (2017-07-15)

• [jsk_perception][people_pose_estimation_2d] publish image only when subscribed (#2164)
• Enhance PeoplePoseEstimation2D (#2162) * Fix run_depend on rviz * Install different pre-trained model according to the version of chainer * Support 16UC1 depth image in PeoplePoseEstimation2D * Visualize people 3D pose on rviz in sample * Add orientation to people 3d pose * Create point cloud in play_rosbag_people.xml * Fix AttributeError of argsort in cupy == 1.0.1
• [jsk_perception][jsk_recognition_utils] support chainer-v2 in alexnet and vgg16 (#2153) * enable alexnet and vgg test * fix syntax in vgg16_object_recognition * alexnet and vgg16 support chainer-v2
• Contributors: Kentaro Wada, Shingo Kitagawa, Yuki Furuta

1.1.3 (2017-07-07)

• [jsk_perception] add FCN-based classifiers (#2142) * make FCN-based classifiers pass test * mask_image_generator run only when use_mask=true * add voc_target_names yaml * FCN-based classifiers publish full result * add sample and test of fcn-based classifiers * add probability_image_classifier node * add label_image_classifier node
• [jsk_perception] squeeze mask to image dim=2 in fcn segmentation (#2144) * check mask ndim before squeeze * add use_mask sample and test for FCN segmentation * fix typo in fcn segmentation * squeeze mask to image dim=2 in fcn segmentation
• [jsk_perception/polygon_to_mask] add error message of frame_id (#2125) * [jsk_perception/polygon_to_mask_image] add error message when frame_id is not correct.
• [jsk_perception] apply candidates node supports topic update (#2143) * node_scripts/apply_context_to_label_probability: update Label msg API * node_scripts/apply_context_to_label_probability: apply candiates support topic update
• [jsk_perception] PeoplePoseEstimation2D (#2115) * [jsk_perception][people_pose_estimation_2d.py] keep compatibility chainer v1 * [jsk_perception/people_pose_estimation_2d] Fixed missed numpy/cupy type * [jsk_perception/people_pose_estimation_2d] Changed sample bag file * [jsk_perception/people_pose_estimation_2d] Add people_mask_publisher * [jsk_perception/people_pose_estimation_2d] Publishe 2d image pose * [jsk_recogntion_msgs/PoseArray] Add score * [jsk_perception/people_pose_estimation_2d] Fixed install sample bag * [jsk_perception/people_pose_estimation_2d] Delete duplicated code * [jsk_perception/people_pose_estimation_2d] Modified type of PeoplePose.msg * [jsk_perception/people_pose_estimation_2d] Fiexed publish img encodings * [jsk_perception/people_pose_estimation_2d] Add test
• [jsk_perception/people_pose] Fixed typo and publish rect images. (#2146 ) * [jsk_perception/people_pose] Refactor. Delete unnecessary code * [jsk_perception/people_pose] Bug fix. Publish rectified image * [jsk_perception/people_pose] Fix typo * [jsk_perception/people_pose] Delete pcl dependencies
• [jsk_perception/draw_rect_array.py] check polygon_msg list size (#2114 )
• [jsk_perception/mask_image_to_rect.cpp] check indices size before execute boundingRect (#2113 ) * [jsk_perception] check indices size before execute boundingRect * jsk_perception/src/mask_image_to_rect.cpp: publish topic even if list is empty
• Contributors: Yuki Furuta, Kanae Kochigami, Masaki Murooka, Shingo Kitagawa, Iori Yanokura

1.1.2 (2017-06-16)

• label_image_decomposer.py: Faster and better visualization of segmentation (#2109 )
• fcn_object_segmentation.{launch,py} : Support .npz in chainermodel (https://github.com/jsk-ros-pkg/jsk_recognition/commit/19d7a2ac09bab2b470a8b06e0ed98d072b4958d4)
• fcn_object_segmentation.{launch,py} : Show deprecated warning for ~model_h5 in fcn_object_segmentation https://github.com/jsk-ros-pkg/jsk_recognition/commit/8d9be278a4ce019f4e026883a30785be874c6a16
• Support chainer v2 in fcn_object_segmentation.py (#2107 )
• tile_image.py : Improve visualization in sample_fuse_depth_image https://github.com/jsk-ros-pkg/jsk_recognition/commit/6caa4c6f5039cb49cf0d07f43a6954a287b8ed35
• Stop using deprecated logging func in jsk_topic_tools (#2097 ) * Stop using deprecated jsk_logxxx
• Refactor cmake to find robot_self_filter (#2089 )
• [jsk_percption][jsk_recogniton_utils] add imagenet_object_recognition launch and its sample (#2085 ) * add sample_imagenet_object_recognition launch * use imagenet launch in alexnet sample launch * add imagenet_object_recognition.launch * move imagenet_target_names in config * install bvlc_vgg16 chainermodel * format API in vgg16: model_h5 -> model_file * format Alex -> AlexNet
• [jsk_perception] add AlexNet object recognition node (#2083 )

• inherit VGG16ObjectRecognition in AlexNet

System Message: WARNING/2 (<string>, line 758)

Bullet list ends without a blank line; unexpected unindent.

• rename alex to alexnet
• mv imagenet_target_names.yaml in sample/config
• add test for alex_object_recognition
• add sample for alex_object_recognition
• add alex_object_recognition node

• jsk_perception/test/bof_histogram_extractor.test: increase time-limit for test_bof_histogram_extractor (#2079)

• fix typo in fcn_object_segmentation (#2076) * Improve the location of squeezing batch axis https://github.com/jsk-ros-pkg/jsk_recognition/commit/ddf46101d2d02e7bd18261542a2bacb456bf6e11

• Remove unexpectedly introduced torch rosdep key (#2074)

• FilterMaskImageWithSize: Filter mask image with its size (#2062) * Add flag of ~use_reference to minimize overhead of synchronizing

  System Message: ERROR/3 (<string>, line 769)

  Unexpected indentation.

  • modified: ../doc/jsk_perception/nodes/filter_mask_image_with_size.md
  • modified: include/jsk_perception/filter_mask_image_with_size.h
  • modified: sample/sample_filter_mask_image_with_size.launch
  • modified: src/filter_mask_image_with_size.cpp

  System Message: WARNING/2 (<string>, line 773)

  Block quote ends without a blank line; unexpected unindent.

  • filter_mask_image_with_size.cpp: Improve rosinfo https://github.com/jsk-ros-pkg/jsk_recognition/commit/5b5455c46f8397d6aa7e1c3d3501e87bf39326ca
  • Add sample, test & doc for FilterMaskImageWithSize https://github.com/jsk-ros-pkg/jsk_recognition/commit/14931792da009ef9468bc1ec3d6419005aca9335 - new file: doc/jsk_perception/nodes/filter_mask_image_with_size.md - new file: doc/jsk_perception/nodes/images/filter_mask_image_with_size.gif - modified: jsk_perception/CMakeLists.txt - new file: jsk_perception/sample/sample_filter_mask_image_with_size.launch - new file: jsk_perception/test/filter_mask_image_with_size.test
  • Filter mask image with its size Modified: - jsk_perception/CMakeLists.txt - jsk_perception/include/jsk_perception/multiply_mask_image.h - jsk_perception/plugins/nodelet/libjsk_perception.xml Added: - jsk_perception/cfg/FilterMaskImageWithSize.cfg - jsk_perception/include/jsk_perception/filter_mask_image_with_size.h - jsk_perception/src/filter_mask_image_with_size.cpp

• Add ~approximate_sync param to ConsensusTracking (#2067) Modified: - doc/jsk_perception/nodes/consensus_tracking.rst - jsk_perception/include/jsk_perception/consensus_tracking.h - jsk_perception/src/consensus_tracking.cpp

• FlowVelocityThresholding: Thresholding with velocity of optical flow (#2060 ) * Add sample/test for FlowVelocityThresholding

  System Message: ERROR/3 (<string>, line 796)

  Unexpected indentation.

  • new file: jsk_perception/nodes/flow_velocity_thresholding.md
  • new file: jsk_perception/nodes/images/flow_velocity_thresholding.gif
  • modified: ../jsk_perception/CMakeLists.txt
  • new file: ../jsk_perception/sample/sample_flow_velocity_thresholding.launch
  • new file: ../jsk_perception/test/flow_velocity_thresholding.test

  System Message: WARNING/2 (<string>, line 801)

  Block quote ends without a blank line; unexpected unindent.

  • Thresholding with velocity of optical flow - modified: CMakeLists.txt - new file: cfg/FlowVelocityThresholding.cfg - new file: include/jsk_perception/flow_velocity_thresholding.h - modified: plugins/nodelet/libjsk_perception.xml - new file: src/flow_velocity_thresholding.cpp

• Generate README by script (#2064 )

• fix typo in fcn_object_segmentation.py (#2063 )

• Add ~queue_size param to MultiplyMaskImage (#2061 ) Modified: - doc/jsk_perception/nodes/multiply_mask_image.md - jsk_perception/src/multiply_mask_image.cpp

• Enhance fcn_object_segmentation.py with PyTorch backend (#2051 ) * Optimization for faster processing

  System Message: ERROR/3 (<string>, line 815)

  Unexpected indentation.

  • modified: jsk_perception/node_scripts/fcn_object_segmentation.py

  System Message: WARNING/2 (<string>, line 816)

  Block quote ends without a blank line; unexpected unindent.

  • Fix api of fcn_object_segmentation.py with PyTorch - modified: jsk_perception/node_scripts/fcn_object_segmentation.py
  • Raise error for unavailable torch & torchfcn
  • Remove install_pytorch.sh
  • Revert "Install packages to devel space" This reverts commit 40e068fc6788087c3a11f914269e93a4538be72e.
  • Fix method
  • Install packages to devel space - new file: install_pytorch.py - deleted: install_pytorch.sh
  • Install PyTorch for CUDA8.0 with rosdep
  • Add instruction of installing torchfcn
  • Remove not needed lines

• [jsk_perception] Add concave_hull_mask_image (#2045 ) * [jsk_perception/concave_hull_mask_image] Fixed header * [jsk_perception/concave_hull_mask_image] Fixed consistency of cfg files * [jsk_perception/concave_hull_mask_image] Fixed max area size * [jsk_perception/concave_hull_mask_image] Fixed cfg for limit of contour area size for inf * [jsk_perception/concave_hull_mask_image] Fixed namespace of filter2D * [jsk_perception/concave_hull_mask_image] Fixed include header lists * [jsk_perception/concave_hull_mask_image] Fixed year

• [jsk_perception/apply_mask_image] Add negative option (#2025 )

• [jsk_perception][detection_interface.l] fix: changing object name affects unexpected side effect (#1974 )

• Contributors: Kei Okada, Kentaro Wada, Shingo Kitagawa, Yuki Furuta, Iory Yanokura

1.1.1 (2017-03-04)

1.1.0 (2017-02-09)

1.0.4 (2017-02-09)

• package.xml: python-chainer -> python-chainer-pip (#2014)
• Contributors: Kentaro Wada

1.0.3 (2017-02-08)

• Fix cpp format of consensus_tracking(#1999)
• Contributors: Kentaro Wada

1.0.2 (2017-01-12)

• fix typo in vgg16_object_recognition (#1990)
• No longer required python-gdown dependency Because python-gdown-pip is installed via jsk_data (#1989)
• Disable bing test on Travis (#1985) Currently the node bing seems not used/changed frequently because it requires opencv3, and I have no time to analyze the unstable test on Travis/Jenkins. That's why I'm disabling it. For #1962
• Contributors: Kei Okada, Kentaro Wada, Shingo Kitagawa

1.0.1 (2016-12-13)

• jsk_perception/node_scripts/speak_when_label_found.py: Speak when target labels are found ( #1923)
• Contributors: Kentaro Wada

1.0.0 (2016-12-12)

• Fix for kinetic build (#1943)

• Add missing packages(jsk_data, opencv_apps) to find_package (#1984)

• Add test & sample

  • calc_flow (#1959)
  • background_subtraction (#1959)
  • mask_image_to_rect (#1961)
  • Add test & sample for grid_label (#1960)
  • Add sample for colorize_float_image (#1956)

• Draw rects on image with PolygonStamped input (#1961)

• sample/sample_rect_array_actual_size_filter.launch : Fix typo of sample data path (#1955)

• colorize_float_image.cpp : Fill black color to nan region (#1956)

• scripts/install_sample_data.py : Fix wrong filename in install_sample_data.py (#1954)

• remove depends to driver_base (#1943)

• Contributors: Kei Okada, Kentaro Wada

0.3.29 (2016-10-30)

• CMakeLists.txt: install nodelet.xml: for get to care about install process in #1929
• Contributors: Kei Okada

0.3.28 (2016-10-29)

• [Major Release] Copy jsk_pcl_ros/srv and jsk_perception/srv files to jsk_recognition_msgs (#1914 )
• Copy deprecated srv files to jsk_recognition_msgs - jsk_pcl_ros/srv -> jsk_recognition_msgs/srv - jsk_perception/srv -> jsk_recognition_msgs/srv TODO - 1. Migrate current code for srv files in jsk_recognition_msgs - 2. Remove srv files in jsk_pcl_ros and jsk_perception
• Contributors: Kei Okada, Kentaro Wada

0.3.27 (2016-10-29)

• Fix rosdep installation for jsk_perception with pip (#1883 ) * Fix pip installation with libleveldb-dev installation

• Publish only masks by split_fore_background.py (#1791 )

  • Stabilize split_fore_background.test
  • Fix nan region as mask 0 region
  • Remove synchronization in split_fore_background.py

• Remove extract_images_sync that merged in image_view (#1633 )

• Remove not used codes: image_saver_sync, publish_header (#1651 )

  • they will be merged in image_view package.
  • for https://github.com/jsk-ros-pkg/jsk_recognition/issues/1648#issuecomment-217344813

• Contributors: Kei Okada, Kentaro Wada

0.3.26 (2016-10-27)

• Stop using deprecated jsk_topic_tools/log_utils.h (#1933)

• Fix unparsable nodelet pluginlib xml file (#1929)

• libcmt: Node to track object on 2D image: ConsensusTracking (#1918)

  • jsk_perception ConsensusTracking depends on libcmt which is not released on hydro
  • libcmt 2.0.17 has been released (#1924)
  • check if header file is installed, before 2.0.17
  • Fix encoding conversion of ROSMsg <-> cv::Mat
  • Add test for consensus_tracking
  • Install sample data for consensus_tracking
  • Add sample of consensus tracking
  • Check window is initialized to start tracking
  • Synchronize polygon and image to set initial tracking window
  • Rename to sample/sample_consensus_tracking.launch
  • Fix coding style of consensus_tracking (follow existing code)
  • Fix year for license
  • Fix name of nodelet of ConsensusTracking
  • Fix place of pkg_check_modules in CMakeLists
  • use package-config version libcmt
  • publish mask image generated from result
  • [jsk_perception] add README and set_rect subscriber which will restart tracking
  • [jsk_perception] add cmt_nodelet depending on libcmt

• Fix for alphabetical order in package.xml (#1908)

• apply_context_to_label_probability: Node to apply context to label probability (#1901) * Add sample for apply_context_to_label_probability * Visualize label_names in label_image_decomposer * Use default GPU=0 in sample_fcn_object_segmentation.launch

  System Message: ERROR/3 (<string>, line 960)

  Unexpected indentation.

  Because it does not work with GPU=-1, CPU mode.

  System Message: WARNING/2 (<string>, line 961)

  Block quote ends without a blank line; unexpected unindent.

  • Apply context to label probability

• Stabilize jsk_perception/sklearn_classifier.test (#1877)

• Stabilize jsk_perception/bing.test (#1877)

• label_image_decomposer.py: Stop using scipy fromimage that is not supported by apt version (#1890)

• Make the test pass (#1897) * Stabilize test for label_image_decomposer * Stabilize test for sklearn_classifer * Stabilize test for bof_histogram_extractor * Comment out unstable test on travis

• Add quality to heightmap (#1886) * [colorize_float_image] fix document and change parameter name. * [jsk_perception, colorize_float_image] fix to handle multi channel image

• fcn_object_segmentation.py: Set bg label for uncertain region of FCN prediction (#1881)

• Contributors: Kei Okada, Kentaro Wada, Yohei Kakiuchi, Yuto Inagaki

0.3.25 (2016-09-16)

0.3.24 (2016-09-15)

• CMakeLists.txt : jsk_data is required in build time, used in scripts/install_sample_data
• Contributors: Kei Okada

0.3.23 (2016-09-14)

• euslisp/eusmodel_template_gen_utils.l: create directory if tepmlate path is not found
• CMakeLists.txt : Makefile.slic is no longer used
• Contributors: Kei Okada

0.3.22 (2016-09-13)

• Basically, if the angle is less than 0, just add 180. Likewise if the angle is greater than 180, just subtract by 180. https://github.com/jsk-ros-pkg/jsk_recognition/pull/1593/files#r77976906
• Sobel operator with higher kernel can give better response https://github.com/jsk-ros-pkg/jsk_recognition/pull/1593#discussion_r77976333
• [jsk_perception] slic as submodule
• sparse_image_encoder.cpp: need to escape %
• remove orientationistogram is not used
• set defiend values to protected member variables
• add doc for image_time_diff.py
• [jsk_perception] Remain executable API for nodes which is moved to opencv_apps Delete deprecated API's cfg and src files.
• Declare jsk_add_rostest in all distros
• Add jsk_ prefix for local macros
• Refactor: jsk_perception_add_rostest -> _add_rostest
• Refactor: jsk_perception -> ${PROJECT_NAME}
• Refactor: jsk_perception_nodelet -> _add_nodelet
• Sort service files
• Fix if block syntax - Use endif() - Use quote "" for VERSION_GREATER
• Fix missing CATKIN_DEPENDS of posedetection_msgs
• Fix node executables installation by introducing macro
• Organize cmake setup order 1. Initialization 2. Download 3. Catkin setup 4. Build 5. Install 6. Test
• Add sample/test for blob_detector (#1849) * Add sample/test for blob_detector * Rename mask image file for understandable name
• Fix special character for double to print (#1836) * Fix special character for double to print * Add unit for percentage in sparse_image_encoder info printing
• Add sample & test for color_histogram node
• Fix image dimension robustness in ExtractImageChannel
• [jsk_perception/src/polygon_to_mask_image.cpp] add warning message when no camera info is available.
• Add test for extract_image_channel.py
• Add sample for extract_image_channel.py
• Extract image channel for channel value in rosparam
• disable global set ssl verification to fase
• Add test for RectArrayToDensityImage
• Add sample for RectArrayToDensityImage
• Add sample for selective_search.py
• Convert rect array to density image
• Publish probability image in fcn_object_segmentation.py
• Publish whole black mask if no contour is found
• Use matplotlib.use('Agg') to make it work on server (without window)
• Update sample/test for drawn label names in label_image_decomposer
• Decompose labels with their names listed as legend
• Test LabelToMaskImage
• Add sample for LabelToMaskImage
• Node to convert label to mask image
• Use std::vector instead of cv::vector for OpenCV3
• Get bounding object mask image from noisy mask image
• replace cv::vector to std::vector
• enable to use cv::vector in opencv-3.x
• Merge pull request #1740 from wkentaro/fcn Fully Convolutional Networks for Object Segmentation
• [jsk_perception/src/virtual_camera_mono.cpp] process only when subscribed
• [jsk_perception/fast_rcnn] Modified avoiding size of rects is 0 case
• Catch error which unexpected size of mask
• Use larger buff_size to process input message with queue_size=1
• Use mask image to enhance the object recognition result
• Use timer and load img file when reconfigured in image_publisher
• Add python-fcn-pip in package.xml
• Add fcn_object_segmentation.launch
• Large size buff_size is required for taking time callback
• Test fcn_object_segmentation.py
• Sample for fcn_object_segmentation.py
• Fully Convolutional Networks for Object Segmentation
• Use small sized image for stable testing
• Make test for sklearn_classifier stable
• Make test for label_image_decomposer stable
• Add sample for slic_super_pixels
• Download trained_data in multiprocess
• Stop drawing boundary on label_image_decomposer - Not so pretty - Maybe Takes time
• Skip when no contours in BoundingRectMaskImage
• Test RectArrayActualSizeFilter
• Add sample for RectArrayActualSizeFilter
• Fix RectArrayActualSizeFilter in terms of size filtering
• Merge pull request #1731 from wkentaro/warn-no-test Warnings for without test node/nodelets
• Merge pull request #1732 from wkentaro/test-with-bof Add test for bof_histogram_extractor.py and sklearn_classifier.py
• jsk_perception/CMakeList.sxt: eigen_INCLUDE_DIRS must be located after catkin_INCLUDE_DIRS
• [jsk_perception] fix bug in solidity_rag_merge
• [polygon_array_color_histogram, polygon_array_color_likelihood] add queue size for message filter
• Warnings for without test node/nodelets
• Add test for bof_histogram_extractor.py and sklearn_classifier.py
• [polygon_array_color_likelihood] add code for reading yaml with latest yaml-cpp
• [jsk_pcl_ros] Fix mistake of rect_array_actual_size_filter
• Add sample for label_image_decomposer and use it in testing
• Add test, sample, and documentation for OverlayImageColorOnMono
• Add dynamic reconfigure for OverlayImageColorOnMono
• Implement OverlayImageColorOnMono
• Merge pull request #1697 from wkentaro/rectify-mask-image Implement ConvexHullMaskImage
• Add sample for mask_image_to_label.py
• Rename publish_fixed_images.launch -> sample_image_publisher.launch
• Use natural name of rqt_gui perspective for bof_object_recognition sample
• Add sample & test for BoundingRectMaskImage
• Implement BoundingRectMaskImage
• Add sample & test for ConvexHullMaskImage
• Implement ConvexHullMaskImage
• Add sample & test for BoundingRectMaskImage
• Implement BoundingRectMaskImage
• Add sample & test for MultiplyMaskImage
• Add sample & test for AddMaskImage
• Fix wrong mask size generated by MaskImageGenerator Fix #1701
• Add sample & test for MaskImageGenerator
• Add sample for apply_mask_image
• Install trained_data all time with dependency on ALL
• Merge pull request #1658 from wkentaro/color_pyx [jsk_recognition_utils] Add label color utility function
• Add test for 'rect_array_to_image_marker.py'
• Use labelcolormap in 'rect_array_to_image_marker.py'
• Use labelcolormap in 'draw_rect_array.py'
• Rename download_trained_data -> install_trained_data.py To follow install_test_data.py.
• Comment out test for vgg16_object_recognition does not work in Jenkins
• Install h5py via rosdep and apt
• Install vgg16 trained model
• Recognize object with VGG16 net
• Rename vgg16 -> vgg16_fast_rcnn
• Fix typo in bof_histogram_extractor.py
• Implement drawing node of classification result
• Rename fast_rcnn_caffenet -> fast_rcnn
• Remove dependency on rbgirshick/fast-rcnn
• CMakeLists.txt: on Hydro contains /opt/ros/hydro/include so we need to add after catkin_INCLUDE_DIRS
• Merge pull request #1627 from wkentaro/use-jsk_data [jsk_perception] Use jsk_data download_data function for test_data
• Merge pull request #1628 from wkentaro/download-jsk_data-trained-data [jsk_perception] Download trained_data with jsk_data function
• Use jsk_data download_data function for test_data
• Download trained_data with jsk_data function
• Add roslaunch_add_file_check with add_rostest
• Comment out bof_object_recognition.test because of no resolved imagesift depends
• Support latest sklearn in BoF feature extraction
• Make jsk_perception depend on imagesift for BoF
• Migrate completely jsk_perception/image_utils.h to jsk_recognition_utils/cv_utils.h
• Stable ros version check by STRGREATER
• Deprecated create_feature0d_dataset.[py,launch] Please use create_sift_dataset.py.
• Make it stable image_cluster_indices_decomposer.test
• Make selective_search.test be stable
• Make slic_super_pixels.test be stable
• Make colorize_float_image.test be stable
• Make colorize_labels test stable
• Make apply_mask_image.test be stable
• Make bof_object_recognition.test stable
• Make kmeans.test be stable
• Make bing.test be stable
• Make jsk_perception depend on image_view2 for ImageMaker2 message
• Fix opencv version condition for bing.test (#1638)
• [jsk_perception] Test tile_image.py (#1635) * Follow name convention sample_tile_image.launch * Test tile_image.py
• Test colorize_float_image (#1636)
• Test mask_image_to_label.py (#1634)
• [jsk_perception] Add test for BoF object recognition sample (#1626) * Refactor: BoF object recognition sample filname * Add test for BoF object recognition sample
• Test apply mask image (#1615) Modified: - jsk_perception/CMakeLists.txt Added: - jsk_perception/test/apply_mask_image.test
• Add rqt_gui perspective file for BoF sample (#1622)
• Test colorize labels (#1614) Modified: - jsk_perception/CMakeLists.txt Added: - jsk_perception/test/colorize_labels.test
• Condition to find OpenCV 3 (> 2.9.9) (#1603)
• Test KMeans (#1612) Modified: - jsk_perception/CMakeLists.txt Added: - jsk_perception/test/kmeans.test
• Compile some nodes only when OpenMP found (#1604)
• Stop passing -z flag to ld with clang (#1602)
• [jsk_perception] Find OpenMP as an optional module (#1600) * Find OpenMP as an optional module * Fix indent of cmake
• Refactoring: Rename test file for consistency (#1611)
• [jsk_perception] Test image_publisher.py (#1613) * Refactoring: remap ~output/camera_info to ~camera_info This is a natural output topic design especially for image_pipeline package. * Test image_publisher.py Added: - jsk_perception/test/image_publisher.test
• [jsk_perception] BING: Binarized Normed Gradients for Objectness Estimation at 300fps (#1598) * Add trained_data/ * Add bing * Download trained_data for bing * Documentation about bing * Add test and sample for bing * Download trained_data for bing automatically
• Add trained_data/ (#1597)
• clf save directory fixed (#1539)
• [jsk_perception/image_cluster_indices_decomposer] fix typo (#1592)
• Contributors: Kei Okada, Kentaro Wada, Kim Heecheol, Masaki Murooka, Ryohei Ueda, Shingo Kitagawa, Shintaro Hori, Yohei Kakiuchi, Yuki Furuta, Iori Yanokura, Hiroto Mizohana

0.3.21 (2016-04-15)

0.3.20 (2016-04-14)

• Add sample/test for image_cluster_indices_decomposer.py (#1580)
• Add sample and test for BoundingBoxToRect (#1577) * Add sample for BoundingBoxToRect Modified: - jsk_perception/CMakeLists.txt Added: - jsk_perception/sample/sample_bounding_box_to_rect.launch - jsk_perception/scripts/install_sample_data.py - jsk_perception/test_data/.gitignore * Add test for BoundingBoxToRect * add an example to the documentation * modified document
• [jsk_perception/bounding_box_to_rect] add rosparam approximate sync and queue_size (#1583) * [jsk_perception/bounding_box_to_rect] add approximate sync and queue_size param * [jsk_perception/bounding_box_to_rect] add parameters in doc
• Visualize ClusterPointIndices for image (#1579)
• Install python executables * Install python executables
• Refactor: Make test filenames consistent
• Fix typo in 'test/test_split_fore_background.test'
• Merge pull request #1568 from wkentaro/draw-rect-array [jsk_perception/draw_rect_array.py] Draw rect_array onto a image
• Add test for jsk_perception/draw_rect_array.py Modified: - jsk_perception/CMakeLists.txt Added: - jsk_perception/test/draw_rect_array.test
• Documentize draw_rect_array.py
• Draw rect_array onto a image Added: - jsk_perception/node_scripts/draw_rect_array.py
• Add example for fast_rcnn_caffenet.py
• Subscribe rect_array as object location proposals
• Test jsk_perception/selective_search.py
• Pass RGB image to dlib.find_candidate_object_locations Modified: - jsk_perception/node_scripts/selective_search.py
• [jsk_perception] include opencv header in rect_array_actual_size_filter.h
• [jsk_perception] Add RectArrayActualSizeFilter Filtering array of rectangle regions based on actual size estimated from depth image.
• Remove duplicated roslint in test_depend Modified: - jsk_perception/package.xml
• Contributors: Kei Okada, Kentaro Wada, Ryohei Ueda, Shingo Kitagawa, Yusuke Niitani

0.3.19 (2016-03-22)

• remove rosbuild from run/build depend
• remove dynamic_reconfigure.parameter_generator, which only used for rosbuild
• Contributors: Kei Okada

0.3.18 (2016-03-21)

• jsk_perception/CMakeLists.txt: remove depends to rosbuild
• Contributors: Kei Okada

0.3.17 (2016-03-20)

• remove dynamic_reconfigure.parameter_generator, which only used for rosbuild
• [jsk_perception] binpack_rect_array.py to enumerate jsk_recognition_msgs/RectArray
• [jsk_perception] Add selective_search.py
• [jsk_perception] Use timer callback to speed up tile_image with no_sync:=true
• [jsk_perception] Cache concatenated image to speed up
• Contributors: Kei Okada, Ryohei Ueda

0.3.16 (2016-02-11)

• Merge pull request #1531 from k-okada/sed_package_xml .travis.yml: sed package.xml to use opencv3
• remove image_view2 from find_package(catkin)
• [jsk_perception/CMakeLists.txt] call one of find_package or pkg_check_modules for robot_self_filter.
• [jsk_perception] Set queue_size=1 for tile_image.py
• [jsk_perception] Fix variable names in edge_detector.cpp
• [jsk_perception] Publish result after initialization
• Contributors: Kei Okada, Masaki Murooka, Ryohei Ueda

0.3.15 (2016-02-09)

• U and V has strange library options; https://github.com/ros/rosdistro/pull/10436#issuecomment-180763393
• [jsk_perception] Do not subscribe camera info in calc_flow
• [jsk_perception] Add more 2d feature samples
• Fix label probabilities output message Modified: - jsk_perception/node_scripts/sklearn_classifier.py
• Add queue_size option for bof_histogram_extractor
• Contributors: Kei Okada, Kentaro Wada, Ryohei Ueda

0.3.14 (2016-02-04)

• Merge pull request #1513 from garaemon/bounding-box-to-rect-array [jsk_perception] BoundingBoxToRectArray and rect_array_to_image_marker.py
• Add ~queue_size option for synchronization Modified: - jsk_perception/include/jsk_perception/apply_mask_image.h - jsk_perception/src/apply_mask_image.cpp
• [jsk_perception/ApplyMask] Add option to clip mask image Modified: - jsk_perception/include/jsk_perception/apply_mask_image.h - jsk_perception/src/apply_mask_image.cpp
• [jsk_perception/tile_image.py] Add ~no_sync parameter to disable synchronization of input topics.
• [jsk_perception] Skip for empty sift features Modified: - jsk_perception/node_scripts/bof_histogram_extractor.py
• [jsk_perception] BoundingBoxToRectArray and rect_array_to_image_marker.py
• [jsk_perception] [kalman-filtered-objectdetection-marker.l] fix code
• added default num_threads_ value and modified readme.md
• Merge branch 'master' of https://github.com/jsk-ros-pkg/jsk_recognition into saliency_map_generator Conflicts: jsk_perception/CMakeLists.txt
• [jsk_perception] Except index error on SolidityRagMerge Modified: - jsk_perception/node_scripts/solidity_rag_merge.py
• parallelized main loop
• [jsk_perception/bof_histogram_extractor.py] Skip if only background image
• [jsk_perception] Skip empty image
• [jsk_perception] Publish info in sample launch file Modified: - jsk_perception/sample/publish_fixed_images.launch
• [jsk_perception] Stop using deprecated PLUGINLIB_DECLARE_CLASS Modified: - jsk_perception/src/color_histogram.cpp - jsk_perception/src/edge_detector.cpp - jsk_perception/src/hough_circles.cpp - jsk_perception/src/sparse_image_decoder.cpp - jsk_perception/src/sparse_image_encoder.cpp
• [jsk_perception] Add solidity_rag_merge This is to find image region with high solidity. Firstly, I will use this for vacuum gripper's approach point decision making. Added: - jsk_perception/node_scripts/solidity_rag_merge.py
• [jsk_perception] Set header correctly Modified: - jsk_perception/node_scripts/label_image_decomposer.py
• Merge pull request #1457 from wkentaro/fix-unconfigured-cmake-packagexml [jsk_perception] Fix unconfigured cmake and manifest
• Merge pull request #1455 from wkentaro/publish-label-fg-bg [jsk_perception] Publish label fg/bg decomposed masks
• [jsk_perception] Check ROS_DISTRO for find_package of robot_self_filter
• [jsk_perception] Fix unconfigured cmake and manifest Modified: - jsk_perception/CMakeLists.txt - jsk_perception/package.xml
• [jsk_perception] Keep original encoding and scale to visualize Modified: - jsk_perception/node_scripts/label_image_decomposer.py
• [jsk_perception] ColorizeLabels info -> debug Modified: - jsk_perception/src/colorize_labels.cpp
• [jsk_perception] Add roslint_cpp not as rostest Modified: jsk_perception/CMakeLists.txt
• [jsk_perception] Publish label fg/bg decomposed masks Modified: - jsk_perception/node_scripts/label_image_decomposer.py
• Merge pull request #1398 from wkentaro/roslint-test-for-node-scripts [jsk_perception] Run roslint for python code
• [jsk_perception] Visualize label in label_image_decomposer.py Modified: - jsk_perception/node_scripts/label_image_decomposer.py
• [jsk_perception] Read reference color histogram from a yaml file in PolygonArrayColorLikelihood to avoid race condition between input topics Modified: - doc/jsk_perception/nodes/polygon_array_color_likelihood.md - jsk_perception/CMakeLists.txt - jsk_perception/include/jsk_perception/polygon_array_color_likelihood.h - jsk_perception/package.xml - jsk_perception/src/polygon_array_color_likelihood.cpp
• [jsk_perception] Keep original resolution if all the input images has same shape and add ~draw_input_topic parameter to draw topic name on the tiled images Modified: - jsk_perception/node_scripts/tile_image.py - jsk_recognition_utils/python/jsk_recognition_utils/visualize.py
• Merge pull request #1426 from wkentaro/merge-sklearn-to-jsk-perception Merge sklearn to jsk_perception
• [jsk_perception] Add basic_2d_features.launch to overview effective technique Added: - jsk_perception/launch/basic_2d_features.launch
• [jsk_perception] Run roslint for python code
• Merge pull request #1438 from wkentaro/image-to-label [jsk_perception] Add image_to_label.py
• [jsk_perception] Use StrictVersions instead of ROS_DISTRO Modified: - jsk_perception/node_scripts/tile_image.py
• [jsk_perception/label_image_decomposer.py] Fix typo Modified: - jsk_perception/node_scripts/label_image_decomposer.py
• [jsk_perception/label_image_decomposer.py] Can specify queue_size Modified: - jsk_perception/node_scripts/label_image_decomposer.py
• [jsk_perception] Fix typo Modified: - jsk_perception/node_scripts/label_image_decomposer.py
• [jsk_perception] Fix tile_image.py for hydro. 1. Disable approximate sync for hydro. it's not supported on hydro 2. Use PIL.Image.frombytes instead of PIL.Image.fromstring
• [jsk_perception] Add image_to_label.py Added: - jsk_perception/node_scripts/image_to_label.py
• [jsk_perception] Fix typo in bof_histogram_extractor.py Modified: - jsk_perception/node_scripts/bof_histogram_extractor.py
• Merge sklearn to jsk_perception Modified: jsk_pcl_ros/CMakeLists.txt jsk_pcl_ros/package.xml jsk_perception/package.xml Added: jsk_perception/node_scripts/random_forest_server.py jsk_perception/sample/random_forest_client_sample.py jsk_perception/sample/random_forest_sample.launch jsk_perception/sample/random_forest_sample_data_x.txt jsk_perception/sample/random_forest_sample_data_y.txt
• added param for printing fps to frame
• nodelet for computing image space saliency map
• Contributors: Kamada Hitoshi, Kei Okada, Kentaro Wada, Ryohei Ueda, Krishneel Chaudhary

0.3.13 (2015-12-19)

0.3.12 (2015-12-19)

• Revert "[jsk_perception] slic as submodule"
• Contributors: Ryohei Ueda

0.3.11 (2015-12-18)

• [jsk_perception] slic as submodule
• Contributors: Ryohei Ueda

0.3.10 (2015-12-17)

• [jsk_perception] Add utils to save images by request or from bagfile I sent PR to upstream: - https://github.com/ros-perception/image_pipeline/pull/159 - https://github.com/ros-perception/image_pipeline/pull/163 - https://github.com/ros-perception/image_pipeline/pull/164 Added: jsk_perception/node_scripts/extract_images_sync jsk_perception/node_scripts/image_saver_sync jsk_perception/node_scripts/publish_header
• [jsk_pcl_ros] Check header.frame_id before resolving 3-D spacially Modified: jsk_pcl_ros/src/multi_plane_extraction_nodelet.cpp jsk_perception/src/polygon_array_color_histogram.cpp jsk_recognition_utils/include/jsk_recognition_utils/pcl_ros_util.h jsk_recognition_utils/src/pcl_ros_util.cpp
• Contributors: Kentaro Wada, Ryohei Ueda

0.3.9 (2015-12-14)

• [jsk_perception] Test slop with test_topic_published.py Depends on https://github.com/jsk-ros-pkg/jsk_common/pull/1254
• [jsk_perception] Specific test name for each test files
• [jsk_perception] test_topic_published.py does not work on hydro travis/jenkins Modified: jsk_perception/CMakeLists.txt
• [jsk_perception] Warn about segfault with large size image in SlicSuperpixel Modified: jsk_perception/src/slic_superpixels.cpp
• [jsk_perception] Test slic_super_pixels
• merge origin/master
• use shared_ptr for self_mask instance.
• Merge remote-tracking branch 'origin/master' into add-robot-mask
• [jsk_perception] Clean up duplicated packages in package.xml
• [jsk_perception] Compute polygon likelihood based on color histogram.
• [jsk_perception] Add PolygonArrayColorHistogram
• add sample launch file.
• add robot_to_mask source files.
• Contributors: Kentaro Wada, Masaki Murooka, Ryohei Ueda

0.3.8 (2015-12-08)

• [jsk_perception] Add CATKIN_ENABLE_TESTING if block
• Use ccache if installed to make it fast to generate object file
• [jsk_perception] Refactor publish_fixed_images.launch and fix test
• [jsk_perception] Test split_fore_background.py
• [jsk_perception] Fix header of split_fore_background
• [jsk_perception] Refactor publish_fixed_images.launch and fix test
• [jsk_perception] Specify encoding by rosparam in image_publisher.py
• [jsk_perception] Refactor image_publisher.py
• [jsk_perception] Fix supported encodings of split_fore_background.py It supports both 16UC1 and 32FC1.
• [jsk_perception] Fix supported encodings of split_fore_background.py It supports both 16UC1 and 32FC1.
• [jsk_perception] Add warnNoRemap in subscribe()
• [split fore background] add conversion for depth image format 32FC1
• [jsk_perception] Set frame_id by rosparam
• [jsk_perception] Publish mask also in SplitForeBackground
• add applying blur to output image on edge detector
• [jsk_perception] Split FG/BG with local depth max
• Contributors: Kei Okada, Kentaro Wada, Shingo Kitagawa, Yohei Kakiuchi

0.3.7 (2015-11-19)

• Use gcc -z defs to check undefined symbols in shared objects (jsk_recognitoin_utils, jsk_pcl_ros, jsk_perception). build_check.cpp cannot run on the environment using multiple processes because of invoking libjsk_pcl_ros.so link.
• Merge pull request #1320 from wkentaro/colorize_labels-with-146-colors [jsk_perception] ColorizeLabels support 20->146 labels
• [jsk_perception] ColorizeLabels support 20->146 labels
• [jsk_perception] Call onInitPostProcess() in last of onInit()
• [jsk_perception] Warn no remapping for input topics
• [jsk_perception] Test whether get topic msg
• [jsk_perception] FastRCNN: (new node)
• [jsk_perception] Test label image decomposer async
• [jsk_perception] Rename SimpleClassifier -> ScikitLearnClassifier
• [jsk_perception] Download trained_data for apc recognition sample
• [jsk_perception] Sort build_depend & run_depend
• [jsk_perception] Publish VectorArray in simple_classifier
• [jsk_perception] Publish VectorArray in bof_histogram_extractor
• [jsk_perception] Convert mask to label image
• [jsk_perception] Convert mask to label image
• [jsk_perception] Make connection based and use ClassificationResult.msg
• [jsk_perception] Care about data size when creating bof data
• [jsk_perception] Specify data size when creating bof data
• [jsk_perception] Update BoF object recognition sample
• [jsk_perception] Extract bof histogram with ConnectionBasedTransport
• [jsk_perception] Create bof & bof_hist dataset
• [jsk_perception] Creating sift dataset script
• [jsk_perception] Move ros node scripts/ -> node_scripts/ Closes #1239
• Merge pull request #1236 from wkentaro/slop-param [jsk_perception] slop as param for label_image_decomposer
• Merge pull request #1235 from wkentaro/skip-0-label-image-decomposer [jsk_perception] Skip 0 label in label_image_decomposer
• [jsk_perception] slop as param for label_image_decomposer
• [jsk_perception] Skip 0 label in label_image_decomposer
• [jsk_perception] Debug output about params
• [jsk_perception] Add LabelImageDecomposer
• [jsk_perception] Rename tile_images -> tile_image
• [jsk_perception] Use ConnectionBasedTransport and get_tile_image()
• [jsk_perception/point_pose_extractor] Remove pragma message in compiling and fix format warning
• add oriented_gradient and oriented_gradient_node to install target and export libraries
• [jsk_perception] Add tile_images.py
• Contributors: Hiroaki Yaguchi, Kei Okada, Kentaro Wada, Ryohei Ueda

0.3.6 (2015-09-11)

0.3.5 (2015-09-09)

0.3.4 (2015-09-07)

• Swap doc soft links (to make 'Edit on GitHub' work)
• ColorizeFloatImage correct image link Closes https://github.com/jsk-ros-pkg/jsk_recognition/issues/1165
• Contributors: Kentaro Wada

0.3.3 (2015-09-06)

• [jsk_perception] README.md -> readthedocs.org
• Revert "[jsk_perception] use sphinx for rosdoc" This reverts commit 9e4ba233599b21c6422ec9a45f395b460c53264d.
• [jsk_perception/TabletopColorDifferenceLikelihood] Use geo/polygon.h instead of geo_util.h
• Contributors: Kentaro Wada, Ryohei Ueda

0.3.2 (2015-09-05)

• [jsk_perception] Ignore autogenerated files
• [jsk_perception] Use histograms to compute distance in TabletopColorDifferenceLikelihood
• Contributors: Ryohei Ueda

0.3.1 (2015-09-04)

• [jsk_pcl_ros, jsk_perception] Fix dependency of jsk_recognition_utils for child packages like jsk_rviz_plugins
• Contributors: Ryohei Ueda

0.3.0 (2015-09-04)

• [jsk_perception/CMakeLists.txt] set ROS_PACKAGE_PATH before run roseus using package://
• [jsk_recognition_utils] Introduce new package jsk_recognition_utils in order to use utility libraries defined in jsk_pcl_ros in jsk_perception
• Contributors: Kei Okada, Ryohei Ueda

0.2.18 (2015-09-04)

• [jsk_perception] Do not specify sexp from cmake, just write in file
• [jsk_perception] Add .gitignore about auto-generated files
• [jsk_perception] Add template directory to run eusmodel_template_gen.l correctly
• [jsk_perception] Add PolygonArrayToLabelImage nodelet
• [jsk_perception] Move matchtemplate.py from src to scripts
• [jsk_perception] Move eusmodel_template_gen.l location from src to euslisp
• [jsk_perception] Do not download trained data in compilation time and add script to donload them
• [jsk_perception] use sphinx for rosdoc
• Revert "[jsk_perception] Add rosdoc.yaml to overwrite default file_patterns"
• [package.xml] Updatae Author
• [jsk_perception] use README.md as mainpage.doc
• [jsk_perception] Add rosdoc.yaml to overwrite default file_patterns
• Contributors: Kei Okada, Kentaro Wada, Ryohei Ueda

0.2.17 (2015-08-21)

0.2.16 (2015-08-19)

• [CMakeLists.txt] we can not use rospack within cmake process
• Contributors: Kei Okada

0.2.15 (2015-08-18)

• Merge pull request #1058 from garaemon/uncomment-generate-template Uncomment generate template
• [jsk_perception] Add executable flag to eusmodel_template_gen.l
• [jsk_perception] uncomment generate template
• Contributors: JSK-PR2, Ryohei Ueda

0.2.14 (2015-08-13)

• [jsk_perception] pub posewithcovariancestamped
• [jsk_perception] Add nodelet ColorizeFloatImage to colorize generic float image
• sliding_window_object_detector : opencv3 has different API for cv::ml::SVM
• src/virtual_camera_mono: use cv.hpp and opencv2 code for cv::getPerspectiveTransform
• src/snake_segmentation: snake (legacy.hpp) is disabled on opencv3
• src/point_pose_extractor: use cv.hpp
• linemode is moved to opencv_contrib, disabled for now (only for opencv3)
• src/calc_flow.cpp: use cv.hpp instead of cv.h
• background_substraction: cv::BackgroundSubtractorMOG2 is abstract type for opencv3
• CMakeLists.txt: depends on cv_bridge, not opencv (jsk_perception)
• [jsk_perception] Update readme
• [jsk_perception] Add simple_classifier*
• [jsk_perception] Scripts for bof and its hist extractor
• do not convert image encode in kmeans and gaussian_blur
• Contributors: Kei Okada, Kentaro Wada, Ryohei Ueda, Hitoshi Kamada, Masaki Murooka

0.2.13 (2015-06-11)

• [jsk_perception] Use dynamic_reconfigure in ImageTimeDiff
• [jsk_perception] Update image_time_diff to use hue/saturation
• [jsk_perception] Add Kmeans section to README
• [jek_perception] Add kmeans
• [jsk_perception] Add GaussignBlur section to README
• [jsk_perception] Add gaussian_blur
• [jsk_perception] Update README.md for squashing dilate/erode
• [jsk_perception] Squash erode/dilate to morphological_operator
• [jsk_perception] Update README.md for morphological operators
• [jsk_perception] Add advanced morphological transformations
• [jsk_perception] Use isBGR/isRGB/isBGRA/isRGBA in ApplyMaskImage
• [jsk_perception] Add isBGR/isRGB/isBGRA/isRGBA
• [jsk_perception] Use header to synchronize in ImageTimeDiff
• [jsk_perception] Update image_time_diff.py to use ImageDifferenceValue.msg
• [jsk_perception] Update docs of image_time_diff for output
• [jsk_perception] Publish with stamp in image_time_diff
• [jsk_perception/image_publisher] Do not exit program even though no file is found
• uncomment camera_info_cb
• add subscription of image_raw
• Updated Sliding window detector. - Removed the trainer - Added Bootstraper
• [jsk_perception] Update README for #927
• [jsk_perception] Enable apply_mask convert mask black to transparent
• Changed from reading saved image from directory to RosBag files
• [jsk_perception] Use jsk_topic_tools/log_utils.h for JSK_ROS_INFO, JSK_NODELET_INFO and so on
• [jsk_perception] add diff per pixel to ImageTimeDiff
• [jsk_perception] Fix bug in apply_mask in converting BGRA/RGBA input image
• [jsk_perception] remove no need get_param in image_publisher
• [jsk_perception] Enable HSVDecomposer to handle BGRA/RGBA image
• [jsk_perception] Enable ApplyMask handle BGRA/RGBA image
• [jsk_perception] ApplyMask Mono8 encoding to publish mask
• [jsk_perception] Add publish_info param to image_publisher
• [jsk_perception] Add dynamic_reconfigure feature to ImagePublisher
• [jsk_perception] Publish the difference between start and current image
• [jsk_perception][ApplyMaskImage] mask image should be mono8
• Node to for training the classifier for Sliding Window Object Detector
• [jsk_perception] Ignore trained_data directory from git filesystem
• Contributors: Kentaro Wada, Ryohei Ueda, Eisoku Kuroiwa, Krishneel Chaudhary

0.2.12 (2015-05-04)

• Revert "[jsk_perception/point_pose_extractor] Use OpenCV's matcher class to estimate mathcing"
• [jsk_perception/point_pose_extractor] Use OpenCV's matcher class to estimate mathcing
• [jsk_perception/point_pose_extractor] Add license header
• [jsk_perception] Untabify point_pose_extractor.cpp
• [jsk_perception/point_pose_extractor] Publish PoseStamped from point_pose_extractor result
• add ROS_INFO
• [jsk_perception] check if pcam.intrinsicMatrix is valid
• [jsk_perception] Download drill trained data in compiling time
• Removed opencv non-free header directive Corrected the nodelet name in CMakeLists.txt
• Corrected the nodelet name in CMakeLists.txt
• Removed opencv non-free header directive
• Nodelet for Edge, Contour Thinning and Nodelet for Sliding window object detector
• [jsk_perception] add Fisheye Rotate parameter
• add upside down option to cfg
• add Fisheye Ray Publisher
• [jsk_perception] Add ProjectImagePoint nodelet to project image local coordinates into 3-D point
• [jsk_perception] Update README for fisheye
• [jsk_perception] update Fisheye To Panoarama
• [jsk_perception] Modify typo
• [jsk_perception] Add MaskImageGenerator
• add scale command to shrink the output and make faster
• add cfg
• [jsk_perception] Add fisheye rectify
• [jsk_perception] Add attributeError message to image_publisher.py
• [jsk_perception] Fix README.md about erode/dilate nodelets
• Merge pull request #834 from wkentaro/update-readme-for-pr-811 [jsk_perception] Update README for histogram max_value of SingleChannelHistogram
• [jsk_perception] Update README for histogram max_value of SingleChannelHistogram
• [jsk_perception] Update README for iterations param of Dilate/ErodeMaskImage
• [jsk_perception] Add iteration param to DilateMaskImage & ErodeMaskImage
• Contributors: Kamada Hitoshi, Kentaro Wada, Ryohei Ueda, Yuto Inagaki, iKrishneel

0.2.11 (2015-04-13)

• add encoded points rate
• Contributors: Kamada Hitoshi

0.2.10 (2015-04-09)

• [jsk_perception] add Simple Fisheye to Panorama
• [jsk_perception] changed order of dynamic reconfigure
• [jsk_perception] default max value of histogram should be 256 to include 255 pixel
• [jsk_perception] print number of point when encoding sparse image
• [jsk_perception] Publish empty camera info from image_publisher.py
• [jsk_perception] Add sample for ColorHistogramLabelMatch
• [jsk_perception] Add documentation about ColorHistogramLabelMatch
• Contributors: Yuki Furuta, Ryohei Ueda, Yuto Inagaki, Kamada Hitoshi, Kentaro Wada

0.2.9 (2015-03-29)

• 0.2.8
• Update Changelog
• Contributors: Ryohei Ueda

0.2.8 (2015-03-29)

0.2.7 (2015-03-26)

0.2.6 (2015-03-25)

0.2.5 (2015-03-17)

• check target cloud data ifnot invalid
• Validate image message without image array (width == 0 and height == 0)
• Enhance: more specific error exception
• Change to avoid SyntaxWarning about not assigning rospy.Publisher argument queue_size
• Change import libs with reasonable order (thirdparty -> ros)
• Contributors: Kentaro Wada, Yu Ohara

0.2.4 (2015-03-08)

• [jsk_perception] Add simple script to publish image file into ros image
• Fix license: WillowGarage -> JSK Lab
• Contributors: Ryohei Ueda

0.2.3 (2015-02-02)

• [jsk_pcl_ros, jsk_perception] Move mask image operation to jsk_perception
• Remove rosbuild files
• [jsk_perception] Add ErodeMaskImage nodelet
• [jsk_perception] Add DilateMaskImage
• Contributors: Ryohei Ueda

0.2.2 (2015-01-30)

• [jsk_perception] add posedetection_msgs
• add image_view2 to depends
• Contributors: Kei Okada

0.2.1 (2015-01-30)

• add image_view2 to depends

0.2.0 (2015-01-29)

0.1.34 (2015-01-29)

• [jsk_perception, checkerboard_detector] Remove dependency to jsk_pcl_ros
• [jsk_pcl_ros, jsk_perception] Move find_object_on_plane from jsk_perception to jsk_pcl_ros to make these packages independent
• [jsk_pcl_ros, jsk_perception] Use jsk_recognition_msgs
• [jsk_pcl_ros, jsk_perception, resized_image_transport] Do not include jsk_topic_tools/nodelet.cmake because it is exported by CFG_EXTRAS
• [imagesift] Better support of masking image: 1) Use jsk_perception::boundingRectOfMaskImage to compute ROI 2) support mask image in imagesift.cpp to make better performance
• [jsk_perception] Export library
• [jsk_perception] Do not use cv::boundingRect to compute bounding box of mask image
• [jsk_perception] install include directory of jsk_perception
• Contributors: Ryohei Ueda

0.1.33 (2015-01-24)

• [jsk_perception] FindObjectOnPlane: Find object on plane from 2d binary image and 3-d polygon coefficients
• [jsk_perception] Publish convex hull image of mask from ContourFinder
• [jsk_perception] Fix min_area parameter to work in BlobDetector
• [jsk_pcl_ros, jsk_perception] Fix CmakeList for catkin build. Check jsk_topic_tools_SOURCE_PREFIX
• [jsk_perception] Add MultiplyMaskImage
• [jsk_perception] Add ~approximate_sync parameter to toggle exact/approximate synchronization
• [jsk_perception] Add UnapplyMaskImage
• [jsk_perception] Add blob image to document
• [jsk_perception] Add BlobDetector
• [jsk_perception] Colorize label 0 as black because label-0 indicates masked region
• [jsk_perception] AddMaskImage to add two mask images into one image
• [jsk_perception] Increase label index of SLICSuperPixels to avoid 0. 0 is planned to be used as 'masked'
• [jsk_perception] Publish result binary image as mono image from ColorHistogramMatch
• [jsk_perception] Extract mask image from coefficients of histogram matching in ColorHistogramLabelMatch
• [jsk_perception] Publish result of coefficient calculation as float image
• [jsk_perception] Support mask image in ColorHistogramLabelMatch
• [jsk_perception] Use OpenCV's function to normalize histogram and add min and max value of histogram in ColorHistogramLabelMatch
• [jsk_perception] Add ~min_value and ~max_value to SingleChannelHistogram
• [jsk_perception] SingleChannelHistogram to compute histogram of single channel image
• [jsk_perception] Add YCrCb decomposer
• [jsk_perception] Add LabDecomposer to decompose BGR/RGB image into Lab color space
• [jsk_perception] Use cv::split to split bgr and hsv image into each channel
• [jsk_perception] Fix metrics of ColorHistogramLabelMatch: 1) correlation original value is [-1:1] and 1 is perfect. we apply (1 - x) / 2 2) chi-squared original value is [0:+inf] and 0 is perfect. we apply 1 / (1 + x^2) 3) intersect original value is [0:1] and 1 is perfect. we apply x 4) bhattacharyya original value is [0:1] and 0 is perfect. we apply 1 - x 5, 6) EMD original value is [0:+inf] and 0 is perfect. we apply 1 / (1 + x^2)
• [jsk_perception] Publish more useful debug image from SLICSuperPixels and add documentation.
• [jsk_perception] Publish image of interest from ColorHistogram
• [jsk_perception] Implement 6 different method to compute coefficients between two histograms
• [jsk_perception] Increase the maximum number of super pixels
• [jsk_perception] Fix ColorHistogram minor bags: 1. Support rect message out side of image 2. Use mask image in HSV histogram calculation
• [jsk_perception] Fix HSVDecomposer color space conversion: support RGB8
• [jsk_perception] color matching based on histogram and label information
• [jsk_perception] Add utlity to visualize mask image: ApplyMaskImage
• [jsk_perception] Add GridLabel
• [jsk_perception] Publish hisotgram messages under private namespace
• [jsk_perception] Add simple launch file as sample of superpixels
• [jsk_perception] Utility to colorize labels of segmentation
• [jsk_perception] Fix SLICSuperPixels: 1) if input image if BGR8 2) transpose the result of clustering
• [jsk_perception] Publish segmentation result as cv::Mat<int> and use patched version of SLIC-SuperPixels to get better performance
• [jsk_perception] Support RGB8 and gray scale color in SLICSuperPixels
• [jsk_perception] Add dynamic_reconfigure interface to SLICSuperPixels
• [jsk_perception] Separate SLICSuperPixels into header and cpp files
• [jsk_perception] Publish result of segmentation of slic superpixels as image
• [jsk_perception] Add snake segmentation
• [jsk_perception] ContourFinder
• [jsk_perception] Support one-channel image in GrabCut
• [jsk_perception] HSVDecomposer to decompose RGB into HSV separate images
• [jsk_perception] Add RGBDecomposer to decompose RGB channels into separate images
• Contributors: Ryohei Ueda

0.1.32 (2015-01-12)

0.1.31 (2015-01-08)

• [jsk_perception] Add parameter to select seed policy (definitely back/foreground or probably back/foreground) to GrabCut
• adapt attention-clipper for fridge demo
• [jsk_perception] Publish mask image of grabcut result
• [jsk_perception] add GrabCut nodelet
• Remove roseus from build dependency of jsk_perception
• added debug pub

0.1.30 (2014-12-24)

0.1.29 (2014-12-24)

• added some more parameters for detection
• Contributors: Yu Ohara

0.1.28 (2014-12-17)

• added param to set threshold of best_Windoq
• Add dynamic reconfigure to background substraction
• Clean up background substraction codes
• Add background substraction
• Support image mask in ColorHistogram
• Separate header and cpp file of color_hisotgram
• Use jsk_topic_tools::DiagnosticNodelet for color histogram
• Fix coding style of color_histogram
• Fix indent of linemod.cpp
• Add linemod sample
• changed color_histogram_matcher to pub box_array defined in jsk_pcl_ros

0.1.27 (2014-12-09)

• added some algolism to get best window
• changed codes to pub center of object
• matchedPointPub by 2dResult of colorhistogram matching
• changed color_histogram_sliding_matcher and added launch to show result
• Contributors: Yu Ohara

0.1.26 (2014-11-23)

0.1.25 (2014-11-21)

• kalmanfilter
• changed name
• added codes in catkin.cmake
• added cfg
• added color_histogram_mathcer_node

0.1.24 (2014-11-15)

• servicecall
• Use intrinsicMatrix instead of projectionMatrix to specify 3x3 matrix(K) instead of 4x3 matrix(P)
• remove eigen and add cmake_modules to find_package for indigo
• fix: use projectionMatrix() for indigo
• Add script to setup training assistant for opencv-like dataset
• Add script to check opencv cascade file
• Script to reject positive data for OpenCV training
• renamed only-perception.launch
• calc existance probability
• removed kalmanlib.l from jsk_perception
• add kalman-filter library
• Contributors: Ryohei Ueda, Hitoshi Kamada, Kei Okada, Kamada Hitoshi

0.1.23 (2014-10-09)

• Install nodelet executables
• mend spell-miss in launch
• modified program to select which camera_info to sub
• renamed camera_node to uvc_camera_node, and added some options
• modified detection-interface.l
• Contributors: Ryohei Ueda, Kamada, Yu Ohara

0.1.22 (2014-09-24)

• Disable ssl when calling git
• Contributors: Ryohei Ueda

0.1.21 (2014-09-20)

• Add more diagnostics to OrganizedMultiPlaneSegmentation and fix global hook for ConvexHull
• Contributors: Ryohei Ueda

0.1.20 (2014-09-17)

0.1.19 (2014-09-15)

0.1.18 (2014-09-13)

• add git to build_depend of jsk_libfreenect2
• Contributors: Ryohei Ueda

0.1.17 (2014-09-07)

• add mk/git to build_depend
• Contributors: Kei Okada

0.1.16 (2014-09-04)

• do not use rosrun in the script of jsk_perception/src/eusmodel_template_gen.sh
• Contributors: Ryohei Ueda

0.1.14 (2014-08-01)

0.1.13 (2014-07-29)

0.1.12 (2014-07-24)

• fix to use catkin to link rospack
• Contributors: Kei Okada, Dave Coleman

0.1.11 (2014-07-08)

• jsk_perception does not depends on pcl, but depends on eigen and tf
• Contributors: Ryohei Ueda

0.1.10 (2014-07-07)

• adding oriented_gradient_node
• add calc_flow program to calc optical flow
• Contributors: Ryohei Ueda, Hiroaki Yaguchi

0.1.9 (2014-07-01)

0.1.8 (2014-06-29)

• initialize _img_ptr at first
• convert color image to GRAY
• add nodelet to detect circles based on hough transformation
• add program to compute color histogram (rgb and hsv color space)
• maked configure_file to create imagesurf, imagestar and imagebrisk automatically
• added the programs to use cv_detection
• Contributors: Ryohei Ueda, Yusuke Furuta, Yu Ohara

0.1.7 (2014-05-31)

0.1.6 (2014-05-30)

0.1.5 (2014-05-29)

• add service interface with sensor_msgs/SetCameraInfo to camshiftdemo, not only mouse selection.
• Contributors: Ryohei Ueda

0.1.4 (2014-04-25)

• add sparse_image program to jsk_percepton
• make edge_detector nodelet class
• Contributors: Ryohei Ueda, Yuki Furuta
• Merge pull request #47 from k-okada/add_rosbuild
• Contributors: Kei Okada

0.1.3 (2014-04-12)

0.1.2 (2014-04-11)

0.1.1 (2014-04-10)

• catkinize jsk_perception
• check initialization in check_subscribers function
• change callback function names for avoiding the same name functions
• add edge_detector.launch
• change debug message
• rename type -> atype
• fix minor bug
• change for treating multiple objects in one ObjectDetection.msg
• add test programs
• add rosbuild_link_boost for compile on fuerte/12.04 , see Issue #224, thanks tnakaoka
• add rectangle_detector, based on http://opencv-code.com/tutorials/automatic-perspective-correction-for-quadrilateral-objects/
• update hoguh_lines
• use blur before canny
• add image_proc modules from opencv samples
• change error_threshold max 200 -> 2000
• add :detection-topic keyword to (check-detection)
• replace sleep to :ros-wait for making interruptible
• add scripts for speaking english
• speak before sleep
• add to spek we're looking for...
• print out debug info
• turtlebot/ros pdf
• add ros/turtlebot-logo images #173
• update japanese speaking
• modify parameter definition. parameter should not be overwritten.
• add option publish-objectdetection-marker
• add slot :diff-rotation in detection_interface.l
• do not create ros::roseus object by load detection_interface.l
• publish tf from sensor frame to detected object pose
• update objectdetection-marker program for new detection_interface
• publish tf and markers, add messages
• print out error value
• fix segfault
• suppor rpy style in relative_pose, status:closed #139
• add :target-object keyword to check-detection
• fix : project3dToPixel was removed in groovy
• update to use cv_bridge
• fix for groovy, use cv_bridge not CvBridge
• fix: speak content
• fix: speak-jp
• fix template location
• add microwave detection sample
• add speak-name for speaking japanease object name
• add speak words
• update detction_interface.l for single detection and speak flag
• add solve-tf parameter for not using tf
• add frame_id for coordinates
• add detection_interface.l for using point_pose_extractor
• remove euclidean_cluster,plane_detector and color_extractor from jsk_perception, they are supported in tabletop and pcl apps should go into jsk_pcl_ros
• add max_output
• add opencv2 to rosdep.yaml for compatibility
• update to fit opencv2 electric/fuerte convention
• fix for fuerte see https://code.ros.org/trac/ros/ticket/3955
• add size check
• fix btVector3 -> tf::Vector3
• fix remove define KdTreePtr
• fix style: support ROSPACK_API_V2 (fuerte)
• support ROSPACK_API_V2 (fuerte)
• fix for pcl > 1.3.0, pcl::KdTree -> pcl::search::KdTree, pcl::KdTreeFLANN -> pcl::search::KdTree
• remove explicit dependency to eigen from jsk_perception
• add whilte_balance_param.yaml
• add publish_array for publishing pointsarray
• move posedetectiondb/SetTemplate -> jsk_perception/SetTemplate
• add color_extractor, plane_detector, euclidean_clustering for jsk_perception
• fixed the package name of WhiteBalance.srv
• add eigen to dependency
• add white_balance_converter to jsk_perception
• change msg from face_detector_mono/Rect -> jsk_perception/Rect. I couldn't find set_serch_rect string under jsk-ros-pkg
• node moved from virtual_camera
• check if the matched region does not too big or too small
• add dynamic reconfigure for point_pose_extractor
• split launch for elevator_navigation, to test modules
• fix for oneiric
• fix for users who does not have roseus in their PATH
• ns can't be empty string in launch xml syntax
• commit updates for demo
• added tv-controller with ut logo
• added tv-controller with ut logo
• fixed the size of wrap image, which is calcurated from input (width/height)
• add to write wrapped image
• add error handling and output template file
• add opencv-logo2.png
• add lipton milktea model, auto generated file prefix .launch -> .xml to avoid listed by auto complete
• add sharp rimokon with ist logo
• changed variable name client -> clients
• add sharp tv controller to sample
• add sample for detection launcher generator
• use try to catch assertions
• set Zero as distortionMatrix, because ImageFeature0D.image is rectified
• fixed the box pose in debug image
• changed code for generate SIFT template info
• use projectionMatrix instead of intrinsicMatrix in solvePnP, remove CvBridge -> cv_bridge
• fix to work without roseus path in PATH
• fix relative pose, object coords to texture coords
• update generation script of SIFT pose estimation launcher, relative pose is not correct
• update eusmodel->sift_perception script
• change detection launch generation script to use jsk_perception/point_pose_extractor
• add std namespace appropriately
• update initialize template method
• publish the debug_image of point_pose_extractor
• chnage the output frame id when using only one template
• change threashold for detectiong object
• use /ObjectDetection_agg instead of /ObjectDetection
• add _agg output topic for debug and logging
• add debug message, set lifetime to 1 sec
• add objectdetection-marker.l
• add relative pose parameter to point_pose_extractor.cpp
• change the PutText region
• update sample launch file, point pose extractor do not subscribe input topics when output is not subscribed
• add viewer_window option to disable the OpenCV window
• empty window name to disable window, point_pose_extractor
• move posedetectiondb to jsk_visioncommon
• moved jsk_vision to jsk_visioncommon
• Contributors: Haseru Chen, Kazuto Murai, Youhei Kakiuchi, Yuki Furuta, Kei Okada, Yuto Inagaki, Manabu Saito, Rosen Dinakov, HiroyukiMikita