利用yolov5和TensorRT从0到1实现目标检测的模型训练到模型部署全过程

Related tags

Deep Learningyolov5-TensorRT
Overview

写在前面

利用TensorRT加速推理速度是以时间换取精度的做法，意味着在推理速度上升的同时将会有精度的下降，不过不用太担心，精度下降微乎其微。此外，要有NVIDIA显卡，经测试，CUDA10.2可以支持20系列显卡及以下，30系列显卡需要CUDA11.x的支持，并且目前有bug。

默认你已经完成了 yolov5的训练过程并得到了.pt模型权值文件。

本文目的仅是带着走通流程。

注意要对应yolov5和tensorrtx的版本。

  • ./yolov5包含yolov5训练以及模型初转化阶段的代码
  • ./model_process是将.wts模型转化为.engine模型的代码
  • ./detector是利用.engine模型进行前向推理阶段的代码

我的运行环境(注意OpenCV要选择适合你的visual studio的版本等问题)：

win10

Visual Studio 2019

NVIDIA GeForce RTX 2060

opencv-3.4.3-vc14_vc15

cuda_10.2.89_441.22_win10

cudnn-10.2-windows10-x64-v7.6.5.32

TensorRT-7.0.0.11.Windows10.x86_64.cuda-10.2.cudnn7.6

cmake-3.21.2-windows-x86_64

上述环境的百度云(测试10、20系列可用)：

链接：https://pan.baidu.com/s/1AyaloTzLap8X2hsJBvyeBw
提取码：dwr7

其他版本下载地址：

CUDA cudnn TensorRT CMake OpenCV

环境安装：

1、安装OpenCV并配置好环境变量

2、安装CUDA

一路默认。一般的安装路径为：C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2

3、安装cudnn和TensorRT

cudnn和TensorRT的安装仅是将下载的对应版本的压缩包解压并复制*.h、*.lib、*.dll到CUDA的安装路径。

1 将cuDNN压缩包解压

2 将cuda\bin中的文件复制到 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin

3 将cuda\include中的文件复制到 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include

4 将cuda\lib中的文件复制到 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\lib

另外，

1 将TensorRT压缩包解压

2 将 TensorRT-7.0.0.11\include中头文件复制到C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include

3 将TensorRT-7.0.0.11\lib中所有lib文件复制到C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\lib\x64

4 将TensorRT-7.0.0.11\lib中所有dll文件复制到C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin

4、安装CMake软件备用

一、将训练阶段得到的.pt模型转化为.wts中间模型

把tensorrtx里面的yolov5\gen_wts.py加入到yolov5里面，执行

python gen_wts.py -w [.pt权值文件路径]

runs\train\exp\weights\best.pt为训练过程生成的.pt模型，生成的best.wts会保存到同目录下，此best.wts待会会用到。

cuda版本每个电脑不一样

配置好的tensorrtx，包括Cmakelist.txt的设定以及dirent.h的配置。

若使用原作者的请参照tensorrtx源码https://github.com/wang-xinyu/tensorrtx ，配置过程中会遇到一些问题，挨个解决，问题不大。

1、在yolov5目录下新建build文件夹

2、修改CMakelist.txt

add_definitions(-DAPI_EXPORTS)

3、打开CMake

​​ generate后关闭

4、yolov5/include/dirent.h

​​ 也可使用我的配置好的

二、利用Cmake软件创建VS工程

修改CMakeLists.txt中此处为你的opencv安装路径。

配置好上方两个目录之后，点击Configure，根据你的环境选择配置，

点击Gnerate，警告可忽视，

现在关闭Cmake即可。

三、wts转化为engine

VS打开刚刚在bulid目录下创建的工程。

build处vs打开，生成

问题:我的模型只识别一个类，需要更改


cd {tensorrtx}/yolov5/

// update CLASS_NUM in yololayer.h if your model is trained on custom dataset

为1

生成项目。

把之前生成的best.wts复制到build\release目录里面

cmd里面运行：

.\test.exe -s .\best.wts best.engine s

运行成功在同文件夹下面会得到best.engine转换后的文件。之后的推理过程使用的都是这个文件。

测试：

.\yolov5.exe -d best.engine .\samples

至此，流程走完。

如果想要进一步封装，可以按照我的示例。

注释掉yolov5.cpp，并取消 几个文件的注释。重新生成项目。按照你的需求更改。

Owner
Helium
Helium
GitHub Repository
Born-Infeld (BI) for AI: Energy-Conserving Descent (ECD) for Optimization

Born-Infeld (BI) for AI: Energy-Conserving Descent (ECD) for Optimization This repository contains the code for the BBI optimizer, introduced in the p

G. Bruno De Luca 5 Sep 06, 2022
Coarse implement of the paper "A Simultaneous Denoising and Dereverberation Framework with Target Decoupling", On DNS-2020 dataset, the DNSMOS of first stage is 3.42 and second stage is 3.47.

SDDNet Coarse implement of the paper "A Simultaneous Denoising and Dereverberation Framework with Target Decoupling", On DNS-2020 dataset, the DNSMOS

Cyril Lv 43 Nov 21, 2022
Pytorch implementation of DeePSiM

Pytorch implementation of DeePSiM

1 Nov 05, 2021
Projects for AI/ML and IoT integration for games and other presented at re:Invent 2021.

Playground4AWS Projects for AI/ML and IoT integration for games and other presented at re:Invent 2021. Architecture Minecraft and Lamps This project i

Vinicius Senger 5 Nov 30, 2022
Customer-Transaction-Analysis - This analysis is based on a synthesised transaction dataset containing 3 months worth of transactions for 100 hypothetical customers.

Customer-Transaction-Analysis - This analysis is based on a synthesised transaction dataset containing 3 months worth of transactions for 100 hypothetical customers. It contains purchases, recurring

Ayodeji Yekeen 1 Jan 01, 2022
[CVPR 2021] NormalFusion: Real-Time Acquisition of Surface Normals for High-Resolution RGB-D Scanning

NormalFusion: Real-Time Acquisition of Surface Normals for High-Resolution RGB-D Scanning Project Page | Paper | Supplemental material #1 | Supplement

KAIST VCLAB 49 Nov 24, 2022
1st Place Solution to ECCV-TAO-2020: Detect and Represent Any Object for Tracking

Instead, two models for appearance modeling are included, together with the open-source BAGS model and the full set of code for inference. With this code, you can achieve around 79 Oct 08, 2022

Residual Pathway Priors for Soft Equivariance Constraints

Residual Pathway Priors for Soft Equivariance Constraints This repo contains the implementation and the experiments for the paper Residual Pathway Pri

Marc Finzi 13 Oct 12, 2022
Using Language Model to Bootstrap Human Activity Recognition Ambient Sensors Based in Smart Homes

Using Language Model to Bootstrap Human Activity Recognition Ambient Sensors Based in Smart Homes This repository is the official implementation of Us

Damien Bouchabou 0 Oct 18, 2021
Some useful blender add-ons for SMPL skeleton's poses and global translation.

Blender add-ons for SMPL skeleton's poses and trans There are two blender add-ons for SMPL skeleton's poses and trans.The first is for making an offli

犹在镜中 154 Jan 04, 2023
Deep Learning tutorials in jupyter notebooks.

DeepSchool.io Sign up here for Udemy Course on Machine Learning (Use code DEEPSCHOOL-MARCH to get 85% off course). Goals Make Deep Learning easier (mi

Sachin Abeywardana 1.8k Dec 28, 2022
SubOmiEmbed: Self-supervised Representation Learning of Multi-omics Data for Cancer Type Classification

SubOmiEmbed: Self-supervised Representation Learning of Multi-omics Data for Cancer Type Classification

Sayed Hashim 3 Nov 15, 2022
Baseline and template code for node21 detection track

Nodule Detection Algorithm This codebase implements a baseline model, Faster R-CNN, for the nodule detection track in NODE21. It contains all necessar

node21challenge 11 Jan 15, 2022
Simple Dynamic Batching Inference

Simple Dynamic Batching Inference 解决了什么问题？ 众所周知，Batch对于GPU上深度学习模型的运行效率影响很大。。。 是在Inference时。搜索、推荐等场景自带比较大的batch，问题不大。但更多场景面临的往往是稀碎的请求(比如图片服务里一次一张图)。 如果

116 Jan 01, 2023
PyTorch implementation of Deformable Convolution

Deformable Convolutional Networks in PyTorch This repo is an implementation of Deformable Convolution. Ported from author's MXNet implementation. Buil

411 Dec 16, 2022
[CVPR 2020] Local Class-Specific and Global Image-Level Generative Adversarial Networks for Semantic-Guided Scene Generation

Contents Local and Global GAN Cross-View Image Translation Semantic Image Synthesis Acknowledgments Related Projects Citation Contributions Collaborat

Hao Tang 131 Dec 07, 2022
This is the official implementation of the paper "Object Propagation via Inter-Frame Attentions for Temporally Stable Video Instance Segmentation".

ObjProp Introduction This is the official implementation of the paper "Object Propagation via Inter-Frame Attentions for Temporally Stable Video Insta

Anirudh S Chakravarthy 6 May 03, 2022
Wordle Env: A Daily Word Environment for Reinforcement Learning

Wordle Env: A Daily Word Environment for Reinforcement Learning Setup Steps: git pull [email&#

2 Mar 28, 2022
Very simple NCHW and NHWC conversion tool for ONNX. Change to the specified input order for each and every input OP. Also, change the channel order of RGB and BGR. Simple Channel Converter for ONNX.

scc4onnx Very simple NCHW and NHWC conversion tool for ONNX. Change to the specified input order for each and every input OP. Also, change the channel

Katsuya Hyodo 16 Dec 22, 2022
[ICML 2022] The official implementation of Graph Stochastic Attention (GSAT).

Graph Stochastic Attention (GSAT) The official implementation of GSAT for our paper: Interpretable and Generalizable Graph Learning via Stochastic Att

85 Nov 27, 2022