Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
liuxinchen3 16554cc92e | 2 years ago | |
---|---|---|
.idea | 2 years ago | |
core | 2 years ago | |
dataset | 2 years ago | |
experiments | 2 years ago | |
models | 2 years ago | |
prepare_files | 2 years ago | |
utils | 2 years ago | |
README.md | 2 years ago | |
augment_lip_sync.py | 2 years ago | |
prepare_files.zip | 2 years ago | |
requirements.txt | 2 years ago | |
search_lip_sync.py | 2 years ago |
Neural Architecture Search for Joint Human Parsing and Pose Estimation (ICCV2021)
prepare_file.zip
to your root path of LIP, such as /home/data/LIP/prepare_file.zip
, and unzip it.pip install -r requirements.txt
We only release the search of interaction (without encoder-decoder search) now.
Searching with 4 gpus:
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nproc_per_node=4 search_lip_sync.py --cfg ./experiments/lip/384_384.yaml
Download the pretrained backbone from link: https://pan.baidu.com/s/1Cs_pf3WsZypwrexrEryS_g
with the extracing code: 0000, and modify the model path in the line 205 of augment_lip_sybc.py.
The newest model weight can be also downloaded from the link.
Training with 4 gpus:
CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nproc_per_node=4 augment_lip_sync.py --cfg ./experiments/lip/384_384.yaml
If you have any questions, please concat with huangyuhang@shu.edu.cn
.
该项目开源了一种使用 NAS 自动搜索人体姿态估计与部位解析联合网络的算法。 该算法使用 NAS 来解决联合学习问题,一方面使用 NAS 来为两个分支搜索特定的编解码 结构来提取各自特征;另一方面,用 NAS 来搜索更高效的交互模块,并设计两种交互模 式,分别用于中间层特征和高层特征的交互,达到更加完全和高效的信息交互。该算法 在两个同时具有人体解析与姿态估计标签的数据集上都取得了业界最好的效果。
Text CSV Python
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》