Home

Awesome

Proactive-Object-Detection

PrObeD:Proactive Object Detection Wrapper

Official Pytorch implementation of Neurips 2023 paper "PrObeD:Proactive Object Detection Wrapper".

Vishal Asnani, Abhinav Kumar, Suya You, Xiaoming Liu

Paper + Supplementary

alt text

Prerequisites

Please look at the environment.yaml file for setting up the environment.

Getting Started

Datasets

Training

Faster R-CNN

python train_faster_rcnn.py --dataset coco --net res101 --bs 8 --nw 1 --lr 0.000001 --lr_decay_step 4 --cuda 

DGNet

python train_dgnet.py --gpu_id 0 --model DGNet

Pre-trained model

The pre-trained model for Faster R-CNN and DGNet can be downloaded from below:

ModelLink
Faster-RCNNModel
DGNetComing soon!!

Testing using pre-trained models

Faster R-CNN

python 14_testnet_2.py --dataset coco --net res101 --cuda --model_path "MODEL PATH" 

DGNet

Run the below command to generate and save the loca;ization maps.

python test_dgnet_loc.py --gpu_id 0 --model DGNet --model_path "MODEL PATH" 

After this, run the evaluation script eval_dgnet.py with the paths of validation data ground-truth and savd predcition to estimate and print all the metrics.

python eval_dgnet.py

Sample pseudo code on how to use our wrapper with any detector

Coming soon!!

If you would like to use our work, please cite:

@inproceedings{asnani2023probed,
  title={PrObeD: Proactive Object Detection Wrapper},
  author={Asnani, Vishal and Kumar, Abhinav and You, Suya and Liu, Xiaoming},
  booktitle={Thirty-seventh Conference on Neural Information Processing Systems},
  year={2023}
}