OpenCV 4.7 版本發(fā)布!算法家族又有新成員
一如既往,OpenCV在新年來臨之際,剛剛更新了小版本,來到了4.7版。
除了繼續(xù)對DNN模塊、算法加速優(yōu)化外,算法家族也有了新成員。
下文來自:
https://github.com/opencv/opencv/wiki/ChangeLog#version470
Highlights of this release:
DNN模塊改進:
New ONNX layers 增加了新的ONNX layers
Signinficant performance optimization for convolutions. Winograd algoritm implementation. 對卷積進行了重要的性能優(yōu)化,實現(xiàn)了卷積優(yōu)化算法Winograd 。
OpenVino 2022.1 support.
Added CANN backend support #22634. Link to the manual: https://github.com/opencv/opencv/wiki/Huawei-CANN-Backend. CANN 是華為開發(fā)的神經(jīng)網(wǎng)絡(luò)加速計算架構(gòu),增加CANN后端支持可以更好支持在華為硬件上的神經(jīng)網(wǎng)絡(luò)推理。
Added batched NMS for multi-class object detection. 增加batched NMS ,用于多類別目標(biāo)檢測。
Algorithms改進與新增:
ArUco markers and April tags support including ChAruco and diamond boards detection and calibration. AR增強現(xiàn)實庫ArUco的進一步支持,ChAruco and diamond boards 的檢測與校正。
QR code detection and decoding quality imrovement. Alignment markers support. Benchmark for QR codes: https://github.com/opencv/opencv_benchmarks/tree/develop/python_benchmarks/qr_codes
QR碼的檢測與校正質(zhì)量改進。性能增強詳見 QR 碼的benchmark。
Nanotrack v2 tracker based on neural networks.
新增基于神經(jīng)網(wǎng)絡(luò)的跟蹤算法Nanotrack v2。
Stackblur algoruthm implementation.
高斯模糊的高效優(yōu)化算法Stackblur的實現(xiàn),“StackBlur作為高斯模糊的近似計算,速度在kernel size <= 5的情況下,接近高斯模糊速度,在大kernel size情況下(如kernel size=101),速度遠(yuǎn)超高斯模糊” 詳見:
https://zhuanlan.zhihu.com/p/574498840
Multimedia 實現(xiàn)優(yōu)化:
FFmpeg 5.x support.
CUDA 12.0 support. Hardware accelerated video codecs support on NVIDIA platforms with modern Video Codec SDK (NVCUVID and NVENCODEAPI).支持CUDA 12.0,支持NVIDIA 平臺的硬件加速編碼。
CV_16UC1 read/write video support with FFmpeg. 支持FFmpeg 用CV_16UC1類型讀寫視頻。
Orientation meta support on Mac with native media API.
New iterator-based API for multi-page image formats.
libSPNG support for PNG format. 支持PNG格式圖像讀寫的libSPNG,安全、易用,且速度更快。
SIMD acceleration for self-built libJPEG-Turbo 對自建的libJPEG-Turbo 進行了SIMD加速。
H264/H265 support on Android. Multiple fixes for video decoder, endcoder and camera memory layout. Android上的H264/H265支持。
G-API
Exposed all core APIs to Python, including stateful kernels. 開發(fā)G-API的Python接口
Optimization:
New universal intrinsics backend for scalable vector instructions. The first scalable implementation for RISC-V RVV 1.0. 支持 RISC-V 指令集。
