No Data
This work is accepted in BMVC2020 as Best Paper Award. It introduces a plugin module in neural network to improve both model accuracy and consistency.
[Project page] | [arXiv] | [Slide] | [Video] | [视频]
I have released beta version Semantic Segmentation code for PASCAL VOC. Please refer to deeplabv3p branch
torch==1.1.0 torchvision==0.2.0
anti-aliasing └── data ├── output ├── ILSVRC2012 └── master └── Adaptive-anti-Aliasing └── ...
| Model Name | | Top-1 Acc | Consistency | |----------------------------------|------------|-----------|-------------| | resnet101k3pasagroup8softmax | weight | 79.0 | 91.8 | | resnet101k5pasagroup8softmax | weight | 78.6 | 92.2 |
python main.py --data ../../data/ILSVRC2012 -f 3 -e -b 32 -a resnet101_pasa_group_softmax --group 8 --weights /pth/to/model
python main.py --data ../../data/ILSVRC2012 -f 3 -b 128 -ba 2 -a resnet101_pasa_group_softmax --group 8 --out-dir /pth/to/output/dir
Please directly put "Adaptive-anti-Aliasing/models_lpf/layers/pasa.py" this module before downsampling layers of the backbone except the first convolution layer. We adopt implemantation directly from:
Instance Segmentation: MaskRcnn
Semantic Segmentation: Deeplab V3+ and TDNet
@inproceedings{zou2020delving, title={Delving Deeper into Anti-aliasing in ConvNets}, author={Xueyan Zou and Fanyi Xiao and Zhiding Yu and Yong Jae Lee}, booktitle={BMVC}, year={2020} }
We borrow most of the code from Richard Zhang's Repo (https://github.com/adobe/antialiased-cnns) Thank you : )