https://github.com/abhaypancholi/activity_detection
This repository consists of the project named "Robust human action detection and acquisition" done by our team. This project aims to recognize and alert to any unusual activity in secured area along with keeping the track id for the person doing the unusual things.
https://github.com/abhaypancholi/activity_detection
computer-vision deep-learning human-activity-detection
Last synced: 12 months ago
JSON representation
This repository consists of the project named "Robust human action detection and acquisition" done by our team. This project aims to recognize and alert to any unusual activity in secured area along with keeping the track id for the person doing the unusual things.
- Host: GitHub
- URL: https://github.com/abhaypancholi/activity_detection
- Owner: AbhayPancholi
- License: mit
- Created: 2023-12-09T08:45:34.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2024-11-10T15:08:35.000Z (over 1 year ago)
- Last Synced: 2024-11-10T16:17:17.310Z (over 1 year ago)
- Topics: computer-vision, deep-learning, human-activity-detection
- Language: Jupyter Notebook
- Homepage:
- Size: 2.07 MB
- Stars: 2
- Watchers: 2
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Robust Human Target Detection and Acquisition
Welcome to the repository for **Robust Human Target Detection and Acquisition**. This project aims to enhance the accuracy of human activity recognition and improve the resilience of human target tracking, especially in complex and occluded security environments.
## Project Overview
This system addresses key challenges in surveillance scenarios:
- **Activity Recognition**: Identifies specific human behaviors like walking, running, or jumping to assess potential security threats.
- **Target Tracking**: Maintains track continuity even when individuals are occluded or temporarily out of the frame.
The project leverages multi-object tracking and re-identification techniques to ensure robust, uninterrupted surveillance. Features such as color histograms from bounding boxes help in tracking individuals despite occlusions, adapting to dynamic environments, lighting changes, and other real-world challenges.
## Key Features
- **Occlusion Handling**: Techniques to mitigate occlusions and ensure consistent tracking.
- **Deep Learning for Activity Recognition**: Utilizes rich feature representations from deep learning to identify activities.
- **Appearance Features**: Enhanced tracking through the extraction of appearance-based features like color and texture.
- **Real-World Adaptability**: Models are designed to perform in various real-world environments, making them suitable for public surveillance, critical infrastructure protection, and security applications.
## Ethical Considerations
We acknowledge the importance of ethical use in surveillance technology. Our project is committed to:
- Privacy preservation and legal compliance.
- Mitigating bias and ensuring transparency in decision-making.
## Future Directions
Looking ahead, the project aims to explore:
- Emerging surveillance technologies.
- Broader applications in security domains.
- Addressing ethical implications and societal impact of widespread surveillance systems.
## Versions
The repository contains two versions of the project, each with its own README file:
- **Version 1**: The older implementation with initial methodologies.
- **Version 2**: The improved version with advanced tracking and activity recognition techniques.