Bhangra Coach
A full-stack dance feedback tool for analyzing Bhangra fundamentals using computer vision and other libraries to compare to a reference, and add feedback.
System Breakdown
Frontend Experience
Upload and review flow
User can sign in, numerous different options to look at progress & history. Then, user uploads a clip, and sees their clip analysis
What it handles
- Upload flow
- Reference selection
- Results UI
Tools used
Backend Pipeline
Processing and comparison
Implements the video processing pipeline using FastAPI and Python. Uploaded clips are processed with MediaPipe to extract pose landmarks frame-by-frame, which are then temporally aligned with a reference sequence to compute movement deltas returned to the frontend as structured feedback.
What it handles
- Processing
- Comparison logic
- Structured results
Tools used
Supabase Layer
Storage and persistence
Stores reference videos and practice videos inputted from the user are also stored here. Supabase is pretty critical for supplying the videos for the users to look at with reference
What it handles
- File storage
- Database
- Sessions
Tools used
Feedback Engine
Movement comparison
Transforms pose comparison into coaching feedback. Furthermore, also displays the reference VERSUS the user's video
What it handles
- Pose signals
- Timing differences
- Readable feedback
Tools used
Gallery
