human-activity-dih18

Human Activity Recognition Using Deep Neural Network

Currently use Neural Network to classify Human Activities We use 3 different types of models cnn2d cnn3d and lstm to classify human activites. Till now 2 different datasets are used. UCF101 and SDHA2010

The project depends on keras, sklearn, opencv, tqdm, requests and their respective dependencies
Use pip install -r requirements.txt to install dependencies.

To train, do the following

1. Edit the config[3d].py file to set the desired hyperparameters and select the DATASET as ucf101 or sdha2010.

2. To download the dataset automatically use the command python dataset.py download

   OR

   Download the dataset manually from http://crcv.ucf.edu/ICCV13-Action-Workshop/download.html

3. Extract the dataset in a folder named videos in the dataset folder.

4. Run python dataset.py extract[3d] for the extraction of frames from the dataset.

5. Run fit[3d].py to train a model from the extracted images

6. All results and vmetrics will be stored in the results folder on completion.

To use 3d networks do the above steps with 3d.py files

For actual live video demonstration we use a stacked yolo model to extract the roi for our models. To use demo,

1. Download the pretrained yolo model into the repo
2. Run python demo.py

Do checkout https://github.com/experiencor/keras-yolo2