# yolov5_tensorrtx_python

**Repository Path**: lllstandout/yolov5_tensorrtx_python

## Basic Information

- **Project Name**: yolov5_tensorrtx_python
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-07-03
- **Last Updated**: 2021-07-03

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# 1 download tensorrtx 

git clone https://github.com/wang-xinyu/tensorrtx 

git reset --hard 4089c64

# 2 convert your  "yolov5s.pt" to yolov5s.wts  ( yolov5 3.0, yolov5 3.1 )
cp tensorrtrtx/gen_wts.py  yolov5-3.1/

cd yolov5-3.1 && python3 gen_wts.py  # output : yolov5s.wts 
# 3 patch tensorrtx 
cp -a  CMakeList.txt python yolov5_lib.cpp yolov5_lib.h tensorrtx/

# 4 build tesorrtx 
# 4.1 modify your net model  INPUT_W , INPUT_H , class_num in yolo_layer.h 
# 4.2 and then 
cd tesorrtx && mkdir build && cd build && cmake ../ && make 

# 5 convert wts to tensorRT 
cp yolov5s.wts tesorrtx/

cd tesorrtx/build  && yolov5 -s  # this will generate yolov5s.engine 
# 6 test 
yolov5 -d img_path 
# 7 build python nodules 
cd tesorrtx/python && python3 setup.py install 

# 8 test python modules 
export LD_LIBRAR_PATH=$PWD/../build  # path for libyolov5_trt.so

./python3 test.py