Hamster Race AI Tracking System
Budget: $250 – $750 USD
Project: Hamster Race Tracker (AI/Computer Vision)
Description:
I need an AI/computer vision solution that can track hamsters running inside a 4-lane oval race track (fixed top-down camera, transparent lane dividers). Each hamster stays in its own lane.
The system should be able to:
Detect each hamster’s position (centroid) in its lane for every video frame.
Assign each hamster to the correct lane (red, blue, green, purple).
Output the direction of movement (clockwise vs counter-clockwise) as well as the position along the track.
Provide clean data in a CSV/JSON format: frame index, lane ID, x/y coordinates, normalized progress along track, direction.
Deliverables:
A Python-based pipeline that:
Takes MP4 race videos as input.
Outputs per-frame tracking data with position + direction.
Includes utility functions for visualizing the hamster’s path (optional overlay).
Instructions/documentation so I can run it on new videos with minimal setup.
Requirements:
Strong background in computer vision & ML (OpenCV, PyTorch/TensorFlow).
Experience with multi-animal tracking frameworks (DeepLabCut, SLEAP, or similar).
Python development skills, ability to deliver a reproducible pipeline.
Nice to Have:
Handling video noise (reflections on transparent walls, motion blur, shadows).
Lightweight solution that doesn’t require massive training datasets.
Description:
I need an AI/computer vision solution that can track hamsters running inside a 4-lane oval race track (fixed top-down camera, transparent lane dividers). Each hamster stays in its own lane.
The system should be able to:
Detect each hamster’s position (centroid) in its lane for every video frame.
Assign each hamster to the correct lane (red, blue, green, purple).
Output the direction of movement (clockwise vs counter-clockwise) as well as the position along the track.
Provide clean data in a CSV/JSON format: frame index, lane ID, x/y coordinates, normalized progress along track, direction.
Deliverables:
A Python-based pipeline that:
Takes MP4 race videos as input.
Outputs per-frame tracking data with position + direction.
Includes utility functions for visualizing the hamster’s path (optional overlay).
Instructions/documentation so I can run it on new videos with minimal setup.
Requirements:
Strong background in computer vision & ML (OpenCV, PyTorch/TensorFlow).
Experience with multi-animal tracking frameworks (DeepLabCut, SLEAP, or similar).
Python development skills, ability to deliver a reproducible pipeline.
Nice to Have:
Handling video noise (reflections on transparent walls, motion blur, shadows).
Lightweight solution that doesn’t require massive training datasets.