OpenCV 4.8 Released With TensorFlow Lite Model Support, AVIF Image Handling

OpenCV 4.8 was released yesterday as the newest feature update to this leading open-source computer vision (CV) library.

OpenCV 4.8 has seen a lot of work to its deep neural network (DNN) module to now support TensorFlow Lite models including INT8 quantized models. There are also various ONNX improvements, various CPU and NVIDIA CUDA fixes, improving the DNN speed on ARM and x86, and full FP16 computation branch support on ARMv8 for a reported 2x the performance of prior releases. The OpenCV 4.8 release also now supports modern Intel OpenVINO releases.

OpenCV 4.8 also has various object detection improvements, calibration fixes, improved RISC-V RVV vector extension support, AVIF image support via libavif within the multimedia code, Python binding improvements, and other changes.

Downloads and more details on the OpenCV 4.8 computer vision library via GitHub and OpenCV.org.