-
[News!] 24-07-01: Our work is accepted by ECCV24. Arxiv Paper can be found here. 🎉
-
[News!] 24-07-12: We build our Project Page which includes a brief summary of our work. 🔥
-
[News!] 24-07-13: We released the Datasets and also Codes. Welcome to use this benchmark and also try our proposed method! 🌟
-
[News!] 24-07-16: We build the leaderboards on Paper With Code: Cross-Domain Few-Shot Object Detection. 🥂
In this paper, we:
- reorganize a benchmark for Cross-Domain Few-Shot Object Detection (CD-FSOD);
- conduct extensive study on several different kinds of detectors (Tab.1 in the paper);
- propose a novel CD-ViTO method via enhancing the existing open-set detector (DE-ViT).
In this repo, we provide:
- links and splits for target datasets;
- codes for our CD-ViTO method;
- codes for the DE-ViT-FT method; (in case you would like to build new methods based on this baseline).
We take COCO as source training data and ArTaxOr, Clipart1k, DIOR, DeepFish, NEU-DET, and UODD as targets.
Also, as stated in the paper, we adopt the "pretrain, finetuning, and testing" pipeline, while the pre-trained stage on COCO is directly taken from the DE-ViT, thus in practice, only the targets are needed to run our experiments.
The target datasets could be easily downloaded in the following links: (If you use the datasets, please cite them properly, thanks.)
An anaconda environment is suggested, take the name "cdfsod" as an example:
git clone git@github.com:lovelyqian/CDFSOD-benchmark.git
conda create -n cdfsod python=3.9
conda activate cdfsod
pip install -r CDFSOD-benchmark/requirements.txt
pip install -e ./CDFSOD-benchmark
cd CDFSOD-benchmark
-
download weights: download pretrained model from DE-ViT.
-
run script:
bash main_results.sh
Add --controller to main_results.sh, then
bash main_results.sh
Our work is built upon DE-ViT, and also we use the codes of ViTDeT, Detic to test them under this new benchmark. Thanks for their work.