arnoldfychen / yolov5_trt

Implement yolov5 s/m//l/x with TensorRT API and support serialization/deserialization+image detection/video detection and recording

yolov5_trt

A implementation of yolov5 s/m//l/x with TensorRT API, it supports serialization/deserialization+image detection/video detection and recording.

This implementation bases on https://github.com/wang-xinyu/tensorrtx/tree/master/yolov5 but made improvement, such as get_rect() and add code for video detection and recording.

Usage:

build out binary:

mkdir build && cd build

cmake ..

make

serialize model to trt engine file

./yolov5 -s [.wts] [.engine] [s/m/l/x or c gd gw]

deserialize out model from trt engine file and do image detection and show result

./yolov5 -d [.engine] [sample_image_dir]

deserialize out model from trt engine file and do video detection and do recording

./yolov5 -v [.engine] video_file