Multi-Person Gaze Tracking & Analytics System
Budget: $250 – $750 USD
Project Title: Real-Time Multi-Person Gaze Tracking & Concentration Analytics
Project Overview
I am building a computer-vision project that can locate and stream the gaze direction of multiple people at once in a live video. The key requirement is real-time performance: as each frame comes in, I need immediate visualization of where every detected face (15-50 people, potentially an entire video call or physical meeting room) is looking.
Technical Requirements
Robustness: The system will be deployed both indoors and outdoors. It must handle glare, shadow shifts, and mixed color temperatures.
Frameworks: OpenCV, MediaPipe, PyTorch, or similar libraries are welcome, provided they maintain low latency and a high frame rate.
Concentration Logic: The system must analyze concentration levels over adjustable intervals (e.g., 1/20 sec to 45 mins). Concentration is defined as "consistently looking forward."
Reporting: At the end of an interval, produce a summary: “Out of 30 people in a 10-minute call, 9 maintained 90% focus, 15 maintained 50%, and 6 showed no focus.”
Hardware Advisory Deliverable
Because this system needs to be deployed in the field, a primary deliverable is a Hardware Specification . You must advise on the best hardware stack to achieve the 15-50 person real-time requirement:
Cameras: Recommend specific sensors or cameras (e.g., Wide FOV, Global Shutter, or IR-capable for low-light/glare robustness).
Processing Units: Advice on edge deployment. Can this run on a Raspberry Pi 5 with an AI Kit, or is an NVIDIA Jetson (Orin/Nano) required? If commodity GPUs are needed, specify minimum VRAM/Cuda core requirements.
Kits: Recommend specific "plug-and-play" kits or enclosures suitable for the indoor/outdoor environment described.
Final Deliverables
Hardware Recommendation Report: Detailed list of suggested cameras, lenses, and processing kits (Raspberry Pi, Jetson, etc.) tailored to this specific use case.
Source Code: Ready to plug into a Python environment with clear setup instructions.
Live Demo Script: Overlays gaze vectors on the video feed in real-time.
Guidelines: Documentation for adjusting thresholds to suit different cameras and lighting conditions.
Qualifications
If you have tackled multi-person gaze or head-pose estimation and can squeeze maximum speed from edge devices or commodity GPUs, your experience is invaluable. In your proposal, please mention:
Which models or landmark detectors you lean on for 15+ faces.
How you plan to handle scaling without the frame rate dropping.
Your initial thoughts on whether a Raspberry Pi can handle this load or if more powerful hardware is mandatory.
Project Overview
I am building a computer-vision project that can locate and stream the gaze direction of multiple people at once in a live video. The key requirement is real-time performance: as each frame comes in, I need immediate visualization of where every detected face (15-50 people, potentially an entire video call or physical meeting room) is looking.
Technical Requirements
Robustness: The system will be deployed both indoors and outdoors. It must handle glare, shadow shifts, and mixed color temperatures.
Frameworks: OpenCV, MediaPipe, PyTorch, or similar libraries are welcome, provided they maintain low latency and a high frame rate.
Concentration Logic: The system must analyze concentration levels over adjustable intervals (e.g., 1/20 sec to 45 mins). Concentration is defined as "consistently looking forward."
Reporting: At the end of an interval, produce a summary: “Out of 30 people in a 10-minute call, 9 maintained 90% focus, 15 maintained 50%, and 6 showed no focus.”
Hardware Advisory Deliverable
Because this system needs to be deployed in the field, a primary deliverable is a Hardware Specification . You must advise on the best hardware stack to achieve the 15-50 person real-time requirement:
Cameras: Recommend specific sensors or cameras (e.g., Wide FOV, Global Shutter, or IR-capable for low-light/glare robustness).
Processing Units: Advice on edge deployment. Can this run on a Raspberry Pi 5 with an AI Kit, or is an NVIDIA Jetson (Orin/Nano) required? If commodity GPUs are needed, specify minimum VRAM/Cuda core requirements.
Kits: Recommend specific "plug-and-play" kits or enclosures suitable for the indoor/outdoor environment described.
Final Deliverables
Hardware Recommendation Report: Detailed list of suggested cameras, lenses, and processing kits (Raspberry Pi, Jetson, etc.) tailored to this specific use case.
Source Code: Ready to plug into a Python environment with clear setup instructions.
Live Demo Script: Overlays gaze vectors on the video feed in real-time.
Guidelines: Documentation for adjusting thresholds to suit different cameras and lighting conditions.
Qualifications
If you have tackled multi-person gaze or head-pose estimation and can squeeze maximum speed from edge devices or commodity GPUs, your experience is invaluable. In your proposal, please mention:
Which models or landmark detectors you lean on for 15+ faces.
How you plan to handle scaling without the frame rate dropping.
Your initial thoughts on whether a Raspberry Pi can handle this load or if more powerful hardware is mandatory.
Related categories:
JavaScript
Python
Algorithm
C++ Programming
OpenCV
Video Processing
Computer Vision
Edge Computing
AI Development