https://github.com/feiyull/tensorrt-alpha

🔥🔥🔥TensorRT for YOLOv8、YOLOv8-Pose、YOLOv8-Seg、YOLOv8-Cls、YOLOv7、YOLOv6、YOLOv5、YOLONAS......🚀🚀🚀CUDA IS ALL YOU NEED.🍎🍎🍎
https://github.com/feiyull/tensorrt-alpha

cuda efficientdet libfacedetection rt-detr tensorrt u2net yolonas yolor yolov3 yolov4 yolov5 yolov6 yolov7 yolov8 yolov8-pose yolov8-seg yolox

Last synced: 10 months ago
JSON representation

🔥🔥🔥TensorRT for YOLOv8、YOLOv8-Pose、YOLOv8-Seg、YOLOv8-Cls、YOLOv7、YOLOv6、YOLOv5、YOLONAS......🚀🚀🚀CUDA IS ALL YOU NEED.🍎🍎🍎

Host: GitHub
URL: https://github.com/feiyull/tensorrt-alpha
Owner: FeiYull
License: gpl-2.0
Created: 2022-09-08T07:13:16.000Z (over 3 years ago)
Default Branch: main
Last Pushed: 2024-07-19T13:17:20.000Z (over 1 year ago)
Last Synced: 2025-04-08T08:17:54.317Z (10 months ago)
Topics: cuda, efficientdet, libfacedetection, rt-detr, tensorrt, u2net, yolonas, yolor, yolov3, yolov4, yolov5, yolov6, yolov7, yolov8, yolov8-pose, yolov8-seg, yolox
Language: C++
Homepage:
Size: 37 MB
Stars: 1,326
Watchers: 33
Forks: 199
Open Issues: 92
Metadata Files:
- Readme: README.md
- License: LICENSE

Awesome Lists containing this project

README

          # TensorRT-Alpha



  [![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_en.md) | 简体中文


  


  


## 可视化



  

  

  

  


  

  

  

  




## 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：


pth -> trt

coming soon.

pth -> onnx -> trt:

- [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。

## 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

- 2023.12.09  🍁 更新 yolov8-pose

- 2023.12.19  🍉 更新 yolov8-seg

- 2023.12.27  💖 更新 yolonas

## 安装

兼容平台: Windows and Linux. 以下环境已被测过：


Ubuntu18.04

- cuda11.3

- cudnn8.2.0

- gcc7.5.0

- tensorrt8.4.2.4

- opencv3.x or 4.x

- cmake3.10.2

Windows10

- cuda11.3 

- cudnn8.2.0

- visual studio 2017 or 2019 or 2022

- tensorrt8.4.2.4

- opencv3.x or 4.x

创建Python环境(可选）

```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  

```

安装教程：

- [Install For Ubuntu18.04](Install_For_Ubuntu18.04/Install_For_Ubuntu18.04.md)


- [Docker For Linux](docker/README.md)


## 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 /home/feiyull/TensorRT-8.4.2.4)

```

开始编译、运行工程，例如:[yolov8](yolov8/README.md)

## 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：



| 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)| 

|[yolov8-pose](yolov8-pose/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)| 

|[yolov8-seg](yolov8-seg/README.md)| |[weiyun](https://share.weiyun.com/3T3mZKBm)| [google driver](https://drive.google.com/drive/folders/1-8phZHkx_Z274UVqgw6Ma-6u5AKmqCOv?usp=sharing)| 

|[yolonas](yolonas/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!)    |      |          |

  

🍉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：



| 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%| 

 	




		

一个batch内，平均每一帧的时间开销	

    
		






## 严格的精度对齐，官方效果 vs TensorRT-Alpha:




			

     	

    

    
		

    yolov8n : Offical( left ) vs Ours( right )	

    
		

    
		

    	

    

    
		

    yolov7-tiny : Offical( left ) vs Ours( right )	

    
		

    
		

    

    
		

    yolov6s : Offical( left ) vs Ours( right )	

    
		

    
		

    

    
		

    yolov5s : Offical( left ) vs Ours( right )	

    
		

    
		

    

    
		

    yolov5s : Offical( left ) vs Ours( right )	

    
		

    
		

    

    
		

    libfacedetection : Offical( left ) vs Ours( right topK:2000)	

    
		

    
		

    



## Citation

```bash

@misc{FeiYull_TensorRT-Alpha,  

  author = {FeiYull},  

  title = {TensorRT-Alpha},  

  year = {2023},  

  publisher = {GitHub},  

  journal = {GitHub repository},  

  howpublished = {https://github.com/FeiYull/tensorrt-alpha}

}

```

ecosyste.ms

Data

Tools

Indexes

Applications

Experiments

Awesome

https://github.com/feiyull/tensorrt-alpha

Awesome Lists containing this project

README