Opened 14 months ago
Closed 12 months ago
New point version.
According to Arch, there was an ABI break with opencv-4.5.4, but no soname bump. Upstream refuses to do a soname bump, saying that there is no ABI guarantee with any OpenCV version.
The same thing applies to the recent libvpx update. Anyone who updates libvpx or opencv needs to rebuild the following packages to use the new libraries:
xsoft/graphweb/firefox.xml: <xref linkend="libvpx"/>,
multimedia/videoutils/ffmpeg.xml: <xref linkend="libvpx"/>,
multimedia/videoutils/mplayer.xml: <xref linkend="libvpx"/>,
multimedia/libdriv/gst10-plugins-good.xml: <xref linkend="libvpx"/>,
multimedia/libdriv/xine-lib.xml: <xref linkend="libvpx"/>,
multimedia/videoutils/ffmpeg.xml: <xref linkend="opencv"/>,
multimedia/libdriv/gst10-plugins-bad.xml: <xref linkend="opencv"/> (with additional modules),
Highlights of this release:
GSoC 2021 is over. 11 projects are successful, and most of the results are already merged into OpenCV tree and available in 4.5.4 (in the main repository or in opencv_contrib). Here is the list of merged functionality:
8-bit quantization in DNN Module: #20228 + onnx importer #20535
Improved OpenCV bindings for Julia: opencv_contib#3009
Speech recognition sample: #20291
Optimizing OpenCV DNN for RISC-V: #20287 + #20521
Tutorial for Universal Intrinsics and parallel_for_ for efficient cross-platform algorithm implementation: #20361
DNN module patches:
Improved layers / activations / supported more models:
GRU, CumSum, Max, Min, ExpandDims
Fixed convolution with asymmetric padding
Fixed Unsqueeze (ONNX opset 13)
Fixed several memory access issues in OpenCL kernels
Implement CTC prefix beam search decode for TextRecognitionModel: #20524
Added SoftNMS implementation: #20813
Intel® Inference Engine backend ( OpenVINO™ ):
added support for OpenVINO 2021.4.1 LTS release
added support for models with non-FP32 outputs or for outputs with 1D layout
Introduced generic visitor(s) for the G-API's own implementation of variant<> data structure: #20039.
Introduced a full-featured graph reshape support for the heterogeneous graphs: #20705.
Fixed cv::RMat handling in the regular execution mode: #20593.
Fixed cv::RMat and introduced cv::MediaFrame handling with the desync() operation: #20747.
OpenVINO™ Toolkit backend:
Introduced FP16 data type support in the OpenVINO™ inference backend: #20490.
Fixed issues with incorrect precision and preprocessing options set for imported binary network blobs: #20528;
Fixed issue with possible extra external resources being held during the inference execution, causing hangs at some platforms (e.g. with GStreamer memory in): #20791.
Extended OpenVINO™ inference backend parameters in Python with methods to configure constant input layers and the number of asynchronous infer requests: #20508;
Aligned the compileStreaming() behavior in Python with the C++ version: #20555;
Covered serialization-related methods with Doxygen documentation: #20163;
Added serialization support for graph constant data: #20372;
Added serialization support for the cv::MediaFrame data structure: #20329;
Merged the first bits of the upcoming oneVPL integration for accelerated Media on Windows;
Samples and demos:
Added various graph-level optimizations to the MTCNN demo: #20383, #20474;
Extended the Semantic Segmentation demo with new type of networks (unet-camvid-onnx-0001): #20476;
Fixed various issues in the Python port of the gaze estimation demo: #20510.
Other changes and fixes:
Fixed various issues with ONNX RT inference tests: #20359.
Extended Render fixture with more parameters: #20771.
And many other contributions:
Add DNN-based face detection and face recognition into modules/objdetect: #20422
Restore LineSegmentDetector (LSD) implementation
Python: cv.Mat wrapper over numpy.ndarray is introduced to handle issues with passing of 3D arrays into C++ algorithms: #19091
Python: support OpenCV extension with pure Python modules: #20611
Debugging: Add gdb pretty printer for cv::Mat: #20547
Add Quicklook for Mat on iOS and macOS: #20457
Add generation of new type of radon checkerboard: #20735
If you have NumPy installed, you can ignore the message regarding numpy.distutils not being available
Fixed at 6f75ffc5d76c6c4224642774bddc07c628d3fb09
Powered by Trac 1.5.3.dev0
By Edgewall Software
© 1998-2022 Gerard Beekmans.