model compression and deploy. compression: 1、quantization: quantization-aware-training, 16/8/4/2-bit(dorefa/Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference)、ternary/binary value(twn/bnn/xnor-net); post-training-quantization, 8-bit(tensorrt); 2、 pruning: normal、regular and group convolutional channel pruning; 3、 group convolution structure; 4、batch-normalization folding for quantization. deploy: tensorrt, fp32/fp16/int8(ptq-calibration)、op-adapt(upsample)、dynamic_shape

mobilev2-yolov5s剪枝、蒸馏，支持ncnn，tensorRT部署。ultra-light but better performence！

The easiest way to use deep metric learning in your application. Modular, flexible, and extensible. Written in PyTorch.

text_recognition_toolbox: The reimplementation of a series of classical scene text recognition papers with Pytorch in a uniform way.

A PyTorch Toolbox for Face Recognition

Latex code for making neural networks diagrams