alisubhan5341 / human-action_recognition-using--convlstm

Semester Project for Computer Vision

BEE12-G1-Human-Action_Recognition-Through-Videos-Usman_Jalil

Human Action Recognition Through Videos using Conv LSTMs

Group Members

• Usman Jalil
  • Registration Number: 346172
• Ali Subhan
  • Registration Number: 337505
• Muhammad Abdullah
  • Registration Number: 334656
• Talha Zahid Ch.
  • Registration Number: 346206

Abstract

This project focuses on implementing Human Action Recognition through Videos using Convolutional Long Short-Term Memory networks (Conv LSTMs). The goal is to develop a robust system that can accurately identify and classify human actions in video sequences. The significance of this work lies in its potential applications in various domains, including surveillance, human-computer interaction, and sports analytics.

Methodology

Our approach involves the use of Convolutional LSTMs, which are well-suited for capturing both spatial and temporal features in video data. The project includes the following key steps:

1. Data Collection: Acquiring a diverse dataset of human actions in video format.
2. Data Preprocessing: Cleaning and formatting the data for training.
3. Model Architecture: Designing and implementing a Convolutional LSTM architecture for action recognition.
4. Training: Training the model on the prepared dataset.
5. Evaluation: Assessing the model's performance using appropriate metrics.
6. Usage: Providing a user-friendly interface for utilizing the trained model.

Repository Structure

• /data: Contains the dataset used for training and testing.
• /src: Source code files for the Convolutional LSTM model.
• /docs: Documentation and project-related files.
• /results: Output and results generated during the project.

Getting Started

To get started with the project, follow these steps:

1. Clone the Repository
2. pip install streamlit watchdog pytube
3. streamlit run app.py
4. To run in detach mode 'nohup streamlit run app.py &'

License

This project is licensed under the MIT License - see the LICENSE file for details.