Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
You can not select more than 25 topics
Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
langdu
8f31209808
|
2 years ago | |
|---|---|---|
| .. | ||
| configs | 2 years ago | |
| README.md | 2 years ago | |
| README_CN.md | 2 years ago | |
| deeplabv3p.yml | 2 years ago | |
| hrnet.yml | 2 years ago | |
| hrnet48.yml | 2 years ago | |
| run_all.sh | 2 years ago | |
| run_benchmark.sh | 2 years ago | |
| run_fp16.sh | 2 years ago | |
| run_fp32.sh | 2 years ago | |
README.md
English | 简体中文
Benchmark Introduction
The content is as follow:
├── README.md
├── configs
│ ├── cityscapes_30imgs.yml
│ ├── fastscnn.yml
│ ├── ocrnet_hrnetw48.yml
│ └── segformer_b0.yml
├── deeplabv3p.yml
├── hrnet.yml
├── hrnet48.yml
├── run_all.sh
├── run_benchmark.sh
├── run_fp16.sh
└── run_fp32.sh
Environment
Use Docker to configure the environment.
- docker image: paddlepaddle/paddle:latest-gpu-cuda10.1-cudnn7
- CUDA 10.1 + cudnn7
- paddle=2.1.2
- py=37
Test
Training Test
git clone https://github.com/PaddlePaddle/PaddleSeg.git
cd PaddleSeg
bash benchmark/run_all.sh
How to Open Profiling
Add the following parameter when training.
--profiler_options="batch_range=[50, 60]; profile_path=model.profile
基于PaddlePaddle复现实时语义分割算法DDRNet
Python Text Java Shell Markdown other