PLRC

Point-Level Region Contrast for Object Detection Pre-Training

This is a PyTorch implementation of the PLRC paper:

@inproceedings{bai2022point,
  title={Point-Level Region Contrast for Object Detection Pre-Training},
  author={Bai, Yutong and Chen, Xinlei and Kirillov, Alexander and Yuille, Alan and Berg, Alexander C},
  booktitle={CVPR},
  year={2022}
}

Preparation

Install PyTorch and ImageNet dataset following the official PyTorch ImageNet training code.

Unsupervised Training

This implementation only supports multi-gpu, DistributedDataParallel training, which is faster and simpler; single-gpu or DataParallel training is not supported.

To do unsupervised pre-training of a ResNet-50 model on ImageNet in an 8-gpu machine, run:

python main_plrc.py \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 

This script uses all the default hyper-parameters as described in the PRLC paper.

Models

Our pre-trained ResNet-50 model and finetuned checkpoints on object detection can be downloaded as following:

| | Pretrained Model | Epoch | | ----------- | :-------------------------------------------------------------------------------------------------: | :------: | Res50 | download link | 100 |

| | Finetuned Model | AP | AP50 | AP75 | | ----------- | :-------------------------------------------------------------------------------------------------: | :------: | :--------: | :--------: | | Res50 | download link | 58.2 | 82.7 | 65.1 |

The APs on Pascal VOC is averaged over 5 times.

Detection

Same as MoCo for object detection transfer, please see moco/detection.

Visualization

For model visualzation, we provide an google colab for better illustration.

License

This project is under the CC-BY-NC 4.0 license. See LICENSE for details.