Stereo Vision Toolkit v1.3.1
Stereo Vision Toolkit v1.3.1
Changes:
- Add ML detector using YOLO or Tensorflow
- Add loading of stereo image pair from image files (.png, .jpg, .jpeg)
- Add toggle to show epipolar lines for checking calibration
- Add RGBD streaming using shared memory (for use with I3DRStereo3D Unity package)
- Add I3DR Titania stereo camera support
- Add Vimba camera integration
- Add colour image capture to Basler cameras
- Add point cloud texture options (depth colourmap or image)
- Add colour to loading/saving stereo videos
- Add about dialogue to show software version
- Add separate camera capture and processing threads
- Add video saving of individual camera images or disparity
- Add WLS filter to OpenCV BM and SGBM matchers
- Add hostname and hostID to license message if I3DRSGM license is not found
- Add new camera serials
- Change load/save YAML calibration to use ROS perception supported YAML (changed from OpenCV FileStorage yaml)
- Upgrade to OpenCV v4.5.0
- Upgrade to CUDA 11.1
- Upgrade to I3DRSGM v1.0.11
- Upgrade to using Pylon v6.1.1.18345 library for Basler cameras
Bug fixes:
- Fix YAML file generates invalid Q matrix
- Fix missing link to online documentation
- Fix overwriting image during save
- Fix non rectified images saved with '_rect' suffix
- Fix maximum point cloud z value being set incorrectly
- Fix camera capture delay when processing load is high
- Fix I3DRSGM failed to initalise when switching to camera with different resolution
Known issues:
- ROS perception YAML's data arrays must be on single line in YAML file.
- WLS filter causes speckle filter to sometimes be disabled.
- Stereo videos are saved uncompressed to avoid lossy compression which would cause issues when loading stereo videos. This will create large file sizes.