software projectCompleted
Signed Language
Real-time gesture recognition using MediaPipe hand/body tracking combined with a custom-trained KNN classifier. Features incremental learning, 3× data augmentation, and temporal dynamics analysis.
- Tracking
- Hands + Face + Pose
- Model
- KNN (k=3)
- Learning
- Incremental
Where it started
~250,000 people in Kurdistan with hearing/speech difficulties need accessible communication tools without expensive hardware.
How I built it
Built a software-only gesture recognizer using MediaPipe (hands, face mesh, pose) and a KNN model with incremental training, confusion tracking, and augmentation — works with any standard webcam.
What it makes possible
Demonstrates that sign language recognition is achievable with zero specialized hardware, using only a laptop camera.
Built with
- AI
- Python
- MediaPipe
- KNN
- scikit-learn