[ICCV2023] Vision HGNN: An Image is More than a Graph of Nodes

By Yan Han, Peihao Wang, Souvik Kundu, Ying Ding and Zhangyang Wang.

Requirements

Pytorch 2.4.1, timm 0.6.13, torchprofile 0.0.4, CUDA-compatible GPU (tested on NVIDIA L40S)

ViHGNN Code

Paper: Vision HGNN: An Image is More than a Graph of Nodes

Training

• Training ViHGNN for 8 GPUs:

python train.py /path/to/imagenet/ --model vihg_ti_224_gelu --num-gpu 8 --sched cosine --epochs 300 --opt adamw -j 8 --warmup-lr 1e-6 --mixup .8 --cutmix 1.0 --model-ema --model-ema-decay 0.99996 --aa rand-m9-mstd0.5-inc1 --color-jitter 0.4 --warmup-epochs 20 --opt-eps 1e-8 --repeated-aug --remode pixel --reprob 0.25 --amp --lr 2e-3 --weight-decay .05 --drop 0 --drop-path .1 -b 128 --output /path/to/save/models/

• Training Pyramid ViHGNN for 8 GPUs:

python train.py /path/to/imagenet/ --model pvihg_ti_224_gelu --num-gpu 8 --sched cosine --epochs 300 --opt adamw -j 8 --warmup-lr 1e-6 --mixup .8 --cutmix 1.0 --model-ema --model-ema-decay 0.99996 --aa rand-m9-mstd0.5-inc1 --color-jitter 0.4 --warmup-epochs 20 --opt-eps 1e-8 --repeated-aug --remode pixel --reprob 0.25 --amp --lr 2e-3 --weight-decay .05 --drop 0 --drop-path .1 -b 128 --output /path/to/save/models/

Evaluation

• Evaluate example:

python train.py /path/to/imagenet/ --model pvihg_ti_224_gelu -b 128 --pretrain_path /path/to/pretrained/model/ --evaluate

Citation

@InProceedings{Han_2023_ICCV,
    author    = {Han, Yan and Wang, Peihao and Kundu, Souvik and Ding, Ying and Wang, Zhangyang},
    title     = {Vision HGNN: An Image is More than a Graph of Nodes},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2023},
    pages     = {19878-19888}
}

Acknowledgement

This repo partially uses code from deep_gcns_torch, timm and ViG.