CVPR2025-DeCafNet

Official Repo for CVPR 2025 Paper -- DeCafNet: Delegate and Conquer for Efficient Temporal Grounding in Long Videos

Installation

1. Set up the environment with conda:

   conda env create -f environment.yml
   conda activate decafnet

2. Install NMS

   cd ./libs/nms
   python setup_nms.py install --user

Evaluation

1. Download model checkpoint and data from here.

2. Update the path in opt.yaml in model checkpoint to point to data path

3. Run evaluation the following command. This will reproduce the result of DeCafNet-30% on Ego4D-NLQ.

python eval.py --name "checkpoint/decafnet_30_nlq" --ckpt "6-36000" 

we will release full training and eval code soon

Citation

@inproceedings{
        Lu2025DeCafNet,
        title={DeCafNet: Delegate and Conquer for Efficient Temporal Grounding in Long Videos},
        author={Zijia Lu and A S M Iftekhar and Gaurav Mittal and Tianjian Meng and Xiawei Wang and Cheng Zhao and Rohith Kukkala and Ehsan Elhamifar and Mei Chen},
        booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
        year={2025},
        }