Team Name:

Team JAM

Team Members:

• Andy Chen
• Michael Sekyi
• Jack Hebert

Project Title:

Using AI Hardware Acceleration to Detect Real-Time Crowd and Traffic Conditions

Key Objective:

• Crowd Density Detection: Develop a robust computer vision model using Hailo 8 on Raspberry Pi to detect crowd density in real-time and categorize it into different levels (e.g., low, medium, high) based on traffic patterns along Engineer's Way.
• Alert System Integration: Create an alert system that can inform users or systems when high-density traffic conditions are detected, potentially integrating with mobile apps or campus information boards.
• Energy Efficiency and Processing Optimization: Leverage the hardware acceleration provided by Hailo 8 to ensure the system runs efficiently, with low latency, minimal energy consumption, and optimized processing for real-time applications.
• Scalability and Deployment Strategy: Design a scalable solution that can easily be deployed across multiple pathways or locations on campus, providing consistent and accurate data on foot traffic.

Technology Stack

• Hardware: Raspberry Pi with Hailo 8 AI processor, camera module (for image capture), Wi-Fi module for data transmission.
• Software: Python, OpenCV, TensorFlow Lite (optimized for edge devices), Hailo SDK for model deployment and optimization, MQTT or WebSocket for data transmission and alert notifications.

Expected Outcomes:

• Real-Time Traffic Analysis Model: A working computer vision model that identifies and categorizes crowd density levels on Engineer's Way with at least 70% accuracy.
• Real-Time Dashboard and Alert System: A dashboard displaying crowd density data with the ability to send real-time alerts based on pre-defined thresholds.
• Low-Power Edge AI Solution: A model optimized to run on the Raspberry Pi + Hailo 8 with minimal power consumption while maintaining real-time performance.
• Scalable Documentation: Detailed project documentation, including setup guides, configuration steps, and recommendations for deploying additional nodes on other pathways

Timeline

Week 1: Planning and Research

Week 2: Data Collection and Model Training

Week 3: Model Optimization and Testing

Week 4: System Integration and Deployment