https://github.com/allrivertosea/yolov5-model-quantization-and-cuda-acceleration
Quantization and deployment of YOLOv5 vehicle and pedestrian 8-object detection model using TensorRT, with CUDA programming for accelerating pre-processing and post-processing.
https://github.com/allrivertosea/yolov5-model-quantization-and-cuda-acceleration
Science Score: 13.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
○.zenodo.json file
-
○DOI references
-
○Academic publication links
-
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (2.3%) to scientific vocabulary
Repository
Quantization and deployment of YOLOv5 vehicle and pedestrian 8-object detection model using TensorRT, with CUDA programming for accelerating pre-processing and post-processing.
Basic Info
- Host: GitHub
- Owner: allrivertosea
- Language: C++
- Default Branch: main
- Size: 16.4 MB
Statistics
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
YOLOv5-model-quantization-and-cuda-acceleration
Quantization and deployment of YOLOv5 vehicle and pedestrian 8-object detection model using TensorRT, with CUDA programming for accelerating pre-processing and post-processing.
项目简介
- 基于Tensorrt and CUDA加速Yolov5.6.0
- 支持Ubuntu20.04
- 支持C++
环境说明
- Tensorrt 8.6.1.6
- Cuda 11.7 Cudnn 8.9.0
- Opencv 4.5.5
- Cmake 3.16.3
- RTX 3090
使用说明
make
cd bin
./od-trt-infer
推理性能
CUDA编程加速说明：前处理resize和BGR2RGB、后处理decode、affine和NMS均编写核函数进行CUDA加速
TensorRT加速说明：模型转换为onnx格式后，量化到INT8精度
前处理和后处理CUDA加速，INT8模型推理加速
前处理和后处理CPU，INT8模型推理加速
Owner
- Login: allrivertosea
- Kind: user
- Repositories: 1
- Profile: https://github.com/allrivertosea
GitHub Events
Total
- Push event: 1
Last Year
- Push event: 1