Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
Octobrist 894f01a55e | 2 years ago | |
---|---|---|
.. | ||
.idea | 2 years ago | |
examples | 2 years ago | |
ntire2021 | 2 years ago | |
setup | 2 years ago | |
.gitignore | 2 years ago | |
LICENSE | 2 years ago | |
README.md | 2 years ago | |
requirements.txt | 2 years ago |
3rd place solution for the NTIRE 2021 Multi-modal Aerial View Object Classification Challenge - Track 1 (SAR) at CVPRW 2021
[Challenge Site]
[Challenge Paper]
[Challenge Award]
Recent advancemenets in deep learning have allowed for efficient and accurate classification of electro-optical (EO) images. However, classification of synthetic-aperature radar (SAR) images lag in accuracy. The objective of the challenge is to classify SAR image chips of vehicles into 10 classes, and this work showcases the strategies used to boost performance.
A center-preserving Cutmix modification coined Central Cutmix is employed alongside traditional flipping and rotations to introduce variation during training.
MobileNet | Cutmix | Central cutmix | Stem stride=1 | Cosine annealing | Accuracy valid data |
Accuracy test data |
---|---|---|---|---|---|---|
V2 | 15.58 | |||||
V2 | x | 16.88 | ||||
V2 | x | 18.05 | ||||
V3-Large | x | 18.96 | ||||
V3-Large | x | x | 21.56 | |||
V3-Large | x | x | x | 22.34 | 26.39 |
cd ./setup/
bash setup.bash
cd ./ntire2021/
python train.py
cd ./ntire2021/
python predict.py
This is an analysis about the evolution of DL model based on the community data of Github. The application scenario of all models is the analysis of synthetic aperture images.
CSV Python Text SVG
MIT
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》