README.md 2.9 KB
Newer Older
W
Wenyu 已提交
1
# Vision Transformer Detection
W
Wenyu 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

## Introduction

- [Context Autoencoder for Self-Supervised Representation Learning](https://arxiv.org/abs/2202.03026)  
- [Benchmarking Detection Transfer Learning with Vision Transformers](https://arxiv.org/pdf/2111.11429.pdf)  

Object detection is a central downstream task used to
test if pre-trained network parameters confer benefits, such
as improved accuracy or training speed. The complexity
of object detection methods can make this benchmarking
non-trivial when new architectures, such as Vision Transformer (ViT) models, arrive.

## Model Zoo

| Backbone | Pretrained | Model | Scheduler | Images/GPU  | Box AP | Config | Download |
|:------:|:--------:|:--------------:|:--------------:|:--------------:|:------:|:------:|:--------:|
W
Wenyu 已提交
18 19
| ViT-base | CAE | Cascade RCNN  | 1x | 1 | 52.7 | [config](./cascade_rcnn_vit_base_hrfpn_cae_1x_coco.yml) | [model](https://bj.bcebos.com/v1/paddledet/models/cascade_rcnn_vit_base_hrfpn_cae_1x_coco.pdparams) |
| ViT-large | CAE | Cascade RCNN  | 1x | 1 | 55.7 | [config](./cascade_rcnn_vit_large_hrfpn_cae_1x_coco.yml) | [model](https://bj.bcebos.com/v1/paddledet/models/cascade_rcnn_vit_large_hrfpn_cae_1x_coco.pdparams) |
W
Wenyu 已提交
20 21

**Notes:**
W
Wenyu 已提交
22 23 24
- Model is trained on COCO train2017 dataset and evaluated on val2017 results of `mAP(IoU=0.5:0.95)
- Base model is trained on 8x32G V100 GPU, large model on 8x80G A100
- The above experiments are based on PaddlePaddle 2.2.2
W
Wenyu 已提交
25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

## Citations
```
@article{chen2022context,
  title={Context autoencoder for self-supervised representation learning},
  author={Chen, Xiaokang and Ding, Mingyu and Wang, Xiaodi and Xin, Ying and Mo, Shentong and Wang, Yunhao and Han, Shumin and Luo, Ping and Zeng, Gang and Wang, Jingdong},
  journal={arXiv preprint arXiv:2202.03026},
  year={2022}
}

@article{DBLP:journals/corr/abs-2111-11429,
  author    = {Yanghao Li and
               Saining Xie and
               Xinlei Chen and
               Piotr Doll{\'{a}}r and
               Kaiming He and
               Ross B. Girshick},
  title     = {Benchmarking Detection Transfer Learning with Vision Transformers},
  journal   = {CoRR},
  volume    = {abs/2111.11429},
  year      = {2021},
  url       = {https://arxiv.org/abs/2111.11429},
  eprinttype = {arXiv},
  eprint    = {2111.11429},
  timestamp = {Fri, 26 Nov 2021 13:48:43 +0100},
  biburl    = {https://dblp.org/rec/journals/corr/abs-2111-11429.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

@article{Cai_2019,
   title={Cascade R-CNN: High Quality Object Detection and Instance Segmentation},
   ISSN={1939-3539},
   url={http://dx.doi.org/10.1109/tpami.2019.2956516},
   DOI={10.1109/tpami.2019.2956516},
   journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
   publisher={Institute of Electrical and Electronics Engineers (IEEE)},
   author={Cai, Zhaowei and Vasconcelos, Nuno},
   year={2019},
   pages={1–1}
}
```