HSIC-based Moving WeightAveraging for Few-Shot Open-Set Object Detection

News

[2024-01-18]：This is the official PyTorch implementation of Foodv1 and Foodv2.

Paper Code

Setup

The code is based on detectron2 v0.3

• Installation

conda create -n Food python=3.8 -y
conda activate Food

• Prepare datasets

You should download：

• train and val set of COCO2017

• trainval and test set of VOC2007、VOC2012

following the structure described below：

datasets/
  coco/
  VOC20{07,12}/

In coco：

coco/
  annotations/
    instances_{train,val}2017.json
    person_keypoints_{train,val}2017.json
  {train,val}2017/

In VOC20{07,12}：

VOC20{07,12}/
  Annotations/
  ImageSets/
    Main/
      trainval.txt
      test.txt
  JPEGImages/

Then we generate all datasets for FOOD:

bash prepare_food_voc_coco.sh

Training and Evaluation

VOC-COCO dataset settings:

bash run_voc_coco_AR.sh

VOC10-5-5 dataset settings:

bash run_voc_AR.sh

Citation

If you find this repo useful, please consider citing our paper:

@inproceedings{foodv2,
  title={HSIC-based Moving Weight Averaging for Few-Shot Open-Set Object Detection},
  author={Binyi Su, Hua Zhang, and Zhong Zhou},
  booktitle={Proceedings of the31st ACM International Conference on Multimedia (MM 23)},
  page={5358--5369},
  year={2023},
  doi={https://doi.org/10.1145/3581783.3611850}
}

@ARTICLE{foodv1,
  author={Binyi Su, Hua Zhang, Jingzhi Li, Zhong Zhou},
  journal={IEEE Transactions on Image Processing}, 
  title={Toward Generalized Few-Shot Open-Set Object Detection}, 
  year={2024},
  volume={33},
  number={},
  pages={1389-1402},
  doi={10.1109/TIP.2024.3364495}}