Home

Awesome

FAC-Net

Foreground-Action Consistency Network for Weakly Supervised Temporal Action Localization<br> Linjiang Huang (CUHK), Liang Wang (CASIA), Hongsheng Li (CUHK)

arXiv ICCV2021

Overview

We argue that existing methods for weakly-supervised temporal activity localization cannot guarantee the foreground-action consistency, that is, the foreground and actions are mutually inclusive. Therefore, we propose a novel method named Foreground-Action Consistency Network (FAC-Net) to address this issue. The experimental results on THUMOS14 are as below.

Method \ mAP(%)@0.1@0.2@0.3@0.4@0.5@0.6@0.7AVG
UntrimmedNet44.437.728.221.113.7---
STPN52.044.735.525.816.99.94.327.0
W-TALC55.249.640.131.122.8-7.6-
AutoLoc--35.829.021.213.45.8-
CleanNet--37.030.923.913.97.1-
MAAN59.850.841.130.620.312.06.931.6
CMCS57.450.841.232.123.115.07.032.4
BM60.456.046.637.526.817.69.036.3
RPN62.357.048.237.227.916.78.136.8
DGAM60.054.246.838.228.819.811.437.0
TSCN63.457.647.837.728.719.410.237.8
EM-MIL59.152.745.536.830.522.716.437.7
BaS-Net58.252.344.636.027.018.610.435.3
A2CL-PT61.256.148.139.030.119.210.637.8
ACM-BANet64.657.748.940.932.321.913.539.9
HAM-Net65.459.050.341.131.020.711.139.8
UM67.561.252.343.433.722.912.141.9
FAC-Net (Ours)67.662.152.644.333.422.512.742.2

Prerequisites

Recommended Environment

Data Preparation

  1. Prepare THUMOS'14 dataset.

    • We recommend using features and annotations provided by this repo.
  2. Place the features and annotations inside a dataset/Thumos14reduced/ folder.

Usage

Training

You can easily train the model by running the provided script.

$ python train_main.py --run-type 0 --model-id 1   # rgb stream
$ python train_main.py --run-type 1 --model-id 2   # flow stream

Make sure you use different model-id for RGB and optical flow. Models are saved in ./ckpt/dataset_name/model_id/

Evaulation

The trained model can be found here. Please change the file name to xxx.pkl (e.g., 100.pkl) and put it into ./ckpt/dataset_name/model_id/. You can evaluate the model referring to the two stream evaluation process.

Single stream evaluation

$ python train_main.py --pretrained --run-type 2 --model-id 1 --load-epoch 100  # rgb stream
$ python train_main.py --pretrained --run-type 3 --model-id 2 --load-epoch 100  # flow stream

load-epoch refers to the epoch of the best model. The best model would not always occur at 100 epoch, please refer to the log in the same folder of saved models to set the load epoch of the best model. Make sure you set the right model-id that corresponds to the model-id during training.

Two stream evaluation

$ python test_main.py --rgb-model-id 1 --flow-model-id 2 --rgb-load-epoch 100 --flow-load-epoch 100

References

We referenced the repos below for the code.

If you find this code useful, please cite our paper.

@InProceedings{Huang_2021_ICCV,
    author    = {Huang, Linjiang and Wang, Liang and Li, Hongsheng},
    title     = {Foreground-Action Consistency Network for Weakly Supervised Temporal Action Localization},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2021},
    pages     = {8002-8011}
}

Contact

If you have any question or comment, please contact the first author of the paper - Linjiang Huang (ljhuang524@gmail.com).