Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
Guanghua Yu 4f96dc2fa8 | 2 years ago | |
---|---|---|
.. | ||
_base_ | 2 years ago | |
README.md | 2 years ago | |
solov2_r50_enhance_coco.yml | 2 years ago | |
solov2_r50_fpn_1x_coco.yml | 3 years ago | |
solov2_r50_fpn_3x_coco.yml | 3 years ago | |
solov2_r101_vd_fpn_3x_coco.yml | 2 years ago |
SOLOv2 (Segmenting Objects by Locations) is a fast instance segmentation framework with strong performance. We reproduced the model of the paper, and improved and optimized the accuracy and speed of the SOLOv2.
Highlights:
solov2_r50_fpn_1x
on Tesla v100 with 8 GPU is only 10 hours.Detector | Backbone | Multi-scale training | Lr schd | Mask APval | V100 FP32(FPS) | GPU | Download | Configs |
---|---|---|---|---|---|---|---|---|
YOLACT++ | R50-FPN | False | 80w iter | 34.1 (test-dev) | 33.5 | Xp | - | - |
CenterMask | R50-FPN | True | 2x | 36.4 | 13.9 | Xp | - | - |
CenterMask | V2-99-FPN | True | 3x | 40.2 | 8.9 | Xp | - | - |
PolarMask | R50-FPN | True | 2x | 30.5 | 9.4 | V100 | - | - |
BlendMask | R50-FPN | True | 3x | 37.8 | 13.5 | V100 | - | - |
SOLOv2 (Paper) | R50-FPN | False | 1x | 34.8 | 18.5 | V100 | - | - |
SOLOv2 (Paper) | X101-DCN-FPN | True | 3x | 42.4 | 5.9 | V100 | - | - |
SOLOv2 | R50-FPN | False | 1x | 35.5 | 21.9 | V100 | model | config |
SOLOv2 | R50-FPN | True | 3x | 38.0 | 21.9 | V100 | model | config |
SOLOv2 | R101vd-FPN | True | 3x | 42.7 | 12.1 | V100 | model | config |
Notes:
mAP(IoU=0.5:0.95)
.Backbone | Input size | Lr schd | V100 FP32(FPS) | Mask APval | Download | Configs |
---|---|---|---|---|---|---|
Light-R50-VD-DCN-FPN | 512 | 3x | 38.6 | 39.0 | model | config |
Optimizing method of enhanced model:
@article{wang2020solov2,
title={SOLOv2: Dynamic, Faster and Stronger},
author={Wang, Xinlong and Zhang, Rufeng and Kong, Tao and Li, Lei and Shen, Chunhua},
journal={arXiv preprint arXiv:2003.10152},
year={2020}
}
PaddleDetection
Python Jupyter Notebook Markdown C++ Java other
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》