Suspicious Activity Recognition

Action Recognition Model to detect Suspicious activities through Surveillance videos. Developed for the project "Armed, Injured and other Suspicious Activity Recognition using Drone Surveillance"¹ as part of the internship at Bennet University (Leadingindia.ai).

Dataset used :

https://www.kaggle.com/mateohervas/dcsass-dataset
Delete the second DCSASS Dataset folder and Labels folder

Requirements :

Python 3.10  
ffmpeg  
opencv-python==4.6.0.66  
keras==2.8.0  
numpy==1.24.2  
pandas==1.4.1  
matplotlib==3.5.1  
scikit-learn==1.0.2  
tensorflow==2.8.0  

Dataset Preprocessing :

1. First extract the dataset folder to the same folder as the repository
2. The utils folder contains 3 python scripts to run
3. run python utilties/resort_dataset.py
4. run python utilties/create_normal_class.py
5. run python utilties/preprocess.py

Training :

The train_set.ipynb contains all functions needed to train model. Run all cells and adjust parameters in main() function for training.
For our experiment we used slowfast model with batch size 8, img_size 224 and frames 25.
Our slowfast model trained on above settings for 100 epochs can be found here https://www.mediafire.com/file/8cupxxdo9kk936c/slowfast_finalmodel.hdf5/file.

Testing :

1. First load model or use existing model after training.
2. Run predictions() giving the video to predict as input. We have provided some sample videos in test/. You can also directly run prediction.py, after manually changing input variable paths.

References :

3DCNN : https://github.com/dipakkr/3d-cnn-action-recognition
Slowfast : https://github.com/facebookresearch/SlowFast
Keras - Slowfast : https://github.com/xuzheyuan624/slowfast-keras
Keras - i3D - https://github.com/dlpbc/keras-kinetics-i3d

Footnotes

1. This is the maintained and original repository for the mirror located here. ↩