TensorRT-Alpha: 🔥🔥🔥TensorRT-Alpha supports YOLOv8、YOLOv7、YOLOv6、YOLOv5、YOLOv4、v3、YOLOX、YOLOR...🚀🚀🚀CUDA IS ALL YOU NEED.🍎🍎🍎It also supports end2end CUDA C acceleration and multi-batch inference.

# TensorRT-Alpha
<div align="center">

[![Cuda](https://img.shields.io/badge/CUDA-11.3-%2376B900?logo=nvidia)](https://developer.nvidia.com/cuda-toolkit-archive)
  [![](https://img.shields.io/badge/TensorRT-8.4.2.4-%2376B900.svg?style=flat&logo=tensorrt)](https://developer.nvidia.com/nvidia-tensorrt-8x-download)
  [![](https://img.shields.io/badge/ubuntu-18.04-orange.svg?style=flat&logo=ubuntu)](https://releases.ubuntu.com/18.04/)
  [![](https://img.shields.io/badge/windows-10-blue.svg?style=flat&logo=windows)](https://www.microsoft.com/)

[English](README.md) | 简体中文<br>

[![](https://img.shields.io/badge/B站-blue.svg?logo=bilibili)](https://www.bilibili.com/video/BV1Ay4y1d7M3/?spm_id_from=333.999.0.0&vd_source=a96c9c3f099f4167807291a34fd50fd5)
  [![](https://img.shields.io/badge/youtube-red.svg?logo=youtube)](https://www.youtube.com/watch?v=e7OcvZcqSVA)
  <br>
  </div>

## 可视化
<div align='center'>
  <img src='.github/facemesh.jpg' width="143px">
  <img src='.github/people.gif' width="338px">
  <img src='.github/yolov8-snow.gif' height="190px" width="230px">
  <br>
  <img src='.github/yolov8-stree.gif'  width="260px">
  <img src='.github/u2net.gif'  width="190px">
  <img src='.github/libfacedet.gif'  width="260px">
  <br>
</div>

## 介绍
本仓库提供深度学习CV领域模型加速部署案例，仓库实现的cuda c支持多batch图像预处理、推理、decode、NMS。大部分模型转换流程为：torch->onnx->tensorrt。
获取onnx文件以下有两种方式：<br>

<details>
<summary>pth -> trt</summary>
coming soon.
</details>

- [i]. 本仓库提供的网盘直接下载onnx。[weiyun](https://share.weiyun.com/3T3mZKBm) or [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)
- [ii]]. 按照本仓库提供的指令，手动从相关源代码框架导出onnx。

</details>

## 更新
- 2023.01.01  🔥 更新 yolov3, yolov4, yolov5, yolov6
- 2023.01.04  🍅 更新 yolov7, yolox, yolor
- 2023.01.05  🎉 更新 u2net, libfacedetection
- 2023.01.08  🚀 全网最快支持yolov8的tensorrt部署
- 2023.01.20     更新 efficientdet, pphunmanseg

## 安装
兼容平台: Windows and Linux. 以下环境已被测过：<br>
<details>
<summary>Ubuntu18.04</summary>

- cuda11.3
- cudnn8.2.0
- gcc7.5.0
- tensorrt8.4.2.4
- opencv3.x or 4.x
- cmake3.10.2
</details>

<details>
<summary>Windows10</summary>

- cuda11.3 
- cudnn8.2.0
- visual studio 2017 or 2019 or 2022
- tensorrt8.4.2.4
- opencv3.x or 4.x
</details>

<details>
<summary>创建Python环境(可选）</summary>

```bash
# install miniconda first
conda create -n tensorrt-alpha python==3.8 -y
conda activate tensorrt-alpha
git clone https://github.com/FeiYull/tensorrt-alpha
cd tensorrt-alpha
pip install -r requirements.txt  
```
</details>

安装教程：
- [Install For Ubuntu18.04](Install_For_Ubuntu18.04/Install_For_Ubuntu18.04.md)<br>
- [图文详细版：Ubuntu18.04环境配置](http://t.csdn.cn/6FmFv)
- docker for linux (waiting for update)<br>
- [图文详细版：Windows10环境配置](http://t.csdn.cn/EwNUU)<br>

## 快速开始
### Ubuntu18.04
设置TensorRT根目录（安装目录）路径:
```bash
git clone https://github.com/FeiYull/tensorrt-alpha
cd tensorrt-alpha/cmake
vim common.cmake
# 把common.cmake文件第20行中的TensorRT_ROOT修改成您的TensorRT安装目录, 例如改成如下:
# set(TensorRT_ROOT /root/TensorRT-8.4.2.4)
```
开始编译、运行工程，例如:[yolov8](yolov8/README.md)

### Windows10
[图文详细版：Windows10环境配置](http://t.csdn.cn/EwNUU)<br>

## 模型
目前已实现30多个主流模型，部分整理好的onnx文件如下列表：
<div align='center'>

| model|tesla v100(32G)|weiyun |google driver |
 :-: | :-: | :-: | :-: |
|[yolov3](yolov3/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|      
|[yolov4](yolov4/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|
|[yolov5](yolov5/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[yolov6](yolov6/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[yolov7](yolov7/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[yolov8](yolov8/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[yolox](yolox/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[yolor](yolor/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[u2net](u2net/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[libfacedetection](libfacedetection/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[facemesh](facemesh/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|     
|[pphumanseg](pphumanseg/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)|  
|[efficientdet](efficientdet/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)| 
|more...(🚀: I will be back soon!)    |      |          |
</div>

🍉稍后在tesla v100 和 A100上测量时间开销!现在看看yolov8n在移动端RTX2070m(8G)的性能表现：
<div align='center'>

| 模型 | 视频分辨率 | 模型输入尺寸 |显存占用 |GPU利用率|
  :-: | :-: | :-: | :-: | :-: |
|yolov8n|1920x1080|8x3x640x640|1093MiB/7982MiB| 14%|

<center>	
<img src=".github/yolov8n-b8-1080p-to-640.jpg"
     alt="无法显示图片时显示的文字"
     style="zoom:40%"/>
<br>		
<center>一个batch内，平均每一帧的时间开销	</center>
    <br>		
</div>
<br>

## 严格的精度对齐，官方效果 vs TensorRT-Alpha:
<br>
<div align='center'>			
     <center>	
    <img src=".github/yolov8n-Offical(left)vsOurs(right).jpg"
         alt="无法显示图片时显示的文字"
         style="zoom:80%"/>
    <br>		
    <center>yolov8n : Offical( left ) vs Ours( right )	</center>
    <br>		
    <br>		
    <center>	
    <img src=".github/yolov7-tiny-Offical(left)vsOurs(right).jpg"
         alt="无法显示图片时显示的文字"
         style="zoom:80%"/>
    <br>		
    <center>yolov7-tiny : Offical( left ) vs Ours( right )	</center>
    <br>		
    <br>		
    <img src=".github/yolov6s-v6.3-Offical(left)vsOurs(right).jpg"
         alt="无法显示图片时显示的文字"
         style="zoom:80%"/>
    <br>		
    <center>yolov6s : Offical( left ) vs Ours( right )	</center>
    <br>		
    <br>		
    <img src=".github/yolov5s-v5.7-Offical(left)vsOurs(right)-img2.jpg"
         alt="无法显示图片时显示的文字"
         style="zoom:80%"/>
    <br>		
    <center>yolov5s : Offical( left ) vs Ours( right )	</center>
    <br>		
    <br>		
    <img src=".github/yolov5s-v5.7-Offical(left)vsOurs(right)-img1.jpg"
         alt="无法显示图片时显示的文字"
         style="zoom:80%"/>
    <br>		
    <center>yolov5s : Offical( left ) vs Ours( right )	</center>
    <br>		
    <br>		
    <img src=".github/libfacedet-Offical(left)vsOurs(right-topk-2000).jpg"
         alt="无法显示图片时显示的文字"
         style="zoom:100%"/>
    <br>		
    <center>libfacedetection : Offical( left ) vs Ours( right topK:2000)	</center>
    <br>		
    <br>		
    </center>
</div>

## 参考
[0].https://github.com/NVIDIA/TensorRT<br>
[1].https://github.com/onnx/onnx-tensorrt<br>
[2].https://github.com/NVIDIA-AI-IOT/torch2trt<br>
[3].https://github.com/shouxieai/tensorRT_Pro<br>
[4].https://github.com/opencv/opencv_zoo<br>

bright-moon-tao/TensorRT-Alpha

简介

发行版

贡献者

近期动态

bright-moon-tao/TensorRT-Alpha .gitee-modal { width: 500px !important; }

简介

发行版

贡献者

近期动态

搜索帮助

bright-moon-tao/TensorRT-Alpha