First, clone the repository locally:
git clone https://github.com/fudan-zvg/SeaFormer
cd SeaFormer/seaformer-cls
Then, install PyTorch 1.6.0+ and torchvision 0.7.0+ and pytorch-image-models 0.3.2:
conda install -c pytorch pytorch torchvision
pip install timm==0.3.2
pip install mmsegmentation
pip install mmcv
Download and extract ImageNet train and val images from http://image-net.org/.
The directory structure is the standard layout for the torchvision datasets.ImageFolder, and the training and validation data is expected to be in the train/ folder and val folder respectively:
/path/to/imagenet/
train/
class1/
img1.jpeg
class2/
img2.jpeg
val/
class1/
img3.jpeg
class/2
img4.jpeg
- SeaFormer on ImageNet-1K
| Method | Size | Acc@1 | #Params (M) | FLOPs (G) | Download |
|---|---|---|---|---|---|
| SeaFormer-Tiny | 224 | 68.1 | 1.8 | 0.1 | [Google] [Baidu] |
| SeaFormer-Small | 224 | 73.4 | 4.1 | 0.2 | [Google] [Baidu] |
| SeaFormer-Base | 224 | 76.4 | 8.7 | 0.3 | [Google] [Baidu] |
| SeaFormer-Large | 224 | 79.9 | 14.0 | 1.2 | [Google] [Baidu] |
The password of Baidu Drive is seaf
To evaluate a pre-trained SeaFormer-Tiny on ImageNet val with a single GPU run:
python validate.py /imagenet/validation/ --model SeaFormer_T --checkpoint /path/to/checkpoint_file
To train SeaFormer_Tiny on ImageNet on a single node with 8 gpus for 600 epochs run:
sh distributed_train.sh 8 /data-path -b 128 --model SeaFormer_T --sched step --epochs 600 --decay-epochs 2.4 --decay-rate .973 --opt rmsproptf --opt-eps .001 -j 7 --warmup-lr 1e-6 --warmup-epochs 10 --weight-decay 2e-5 --drop 0.2 --drop-connect 0.2 --model-ema --model-ema-decay 0.9999 --aa rand-m9-mstd0.5 --remode pixel --reprob 0.2 --mixup 0.0 --amp --lr .064 --lr-noise 0.42 --output /output_dir --experiment SeaFormer_T --resume /resume_dir
To train SeaFormer_Small on ImageNet on a single node with 8 gpus for 600 epochs run:
sh distributed_train.sh 8 /data-path -b 128 --model SeaFormer_S --sched step --epochs 600 --decay-epochs 2.4 --decay-rate .973 --opt rmsproptf --opt-eps .001 -j 7 --warmup-lr 1e-6 --warmup-epochs 40 --weight-decay 2e-5 --drop 0.2 --drop-connect 0.2 --model-ema --model-ema-decay 0.9999 --aa rand-m9-mstd0.5 --remode pixel --reprob 0.2 --mixup 0.2 --amp --lr .064 --lr-noise 0.42 --output /output_dir --experiment SeaFormer_S --resume /resume_dir
To train SeaFormer_Base on ImageNet on a single node with 8 gpus for 600 epochs run:
sh distributed_train.sh 8 /data-path -b 128 --model SeaFormer_B --sched step --epochs 600 --decay-epochs 2.4 --decay-rate .973 --opt rmsproptf --opt-eps .001 -j 7 --warmup-lr 1e-6 --warmup-epochs 10 --weight-decay 2e-5 --drop 0.2 --drop-connect 0.2 --model-ema --model-ema-decay 0.9999 --aa rand-m9-mstd0.5 --remode pixel --reprob 0.2 --mixup 0.2 --amp --lr .064 --lr-noise 0.42 --output /output_dir --experiment SeaFormer_B --resume /resume_dir
To train SeaFormer_Large on ImageNet on a single node with 8 gpus for 600 epochs run:
sh distributed_train.sh 8 /data-path -b 128 --model SeaFormer_L --sched step --epochs 600 --decay-epochs 2.4 --decay-rate .973 --opt rmsproptf --opt-eps .001 -j 7 --warmup-lr 1e-6 --warmup-epochs 10 --weight-decay 2e-5 --drop 0.2 --drop-connect 0.2 --model-ema --model-ema-decay 0.9999 --aa rand-m9-mstd0.5 --remode pixel --reprob 0.2 --mixup 0.2 --amp --lr .064 --lr-noise 0.42 --output /output_dir --experiment SeaFormer_L --resume /resume_dir
This repository is released under the Apache 2.0 license as found in the LICENSE file.