This repository is for Audio-Guided Attention Network for Weakly Supervised Violence Detection (ICCECE 2022). The original paper can be found here. Please contact me by email if you have any further queries.

• Download the extracted I3D features of XD-Violence dataset from here.
• Change the file paths of make_list.py in the list folder to generate the training and test list.
• Change the hyperparameters in option.py, where we keep default settings as mentioned in our paper.
• Run the following command for training:

python main.py

• Change the checkpoint path of infer.py.
• Run the following command for test:

python infer.py

The implementation mainly references the repo of XDVioDet. We greatly appreciate their excellent contribution.