Real-time gesture intelligence
Translate Sign Language in Real Time.
A fast, lightweight pipeline powered by MediaPipe and OpenCV. Capture hand landmarks, train a compact classifier, and run live inference in milliseconds.
Built for speed and clarity
The pipeline keeps the moving parts small so you can iterate on new gestures quickly.
Landmark Precision
MediaPipe extracts 21 keypoints per hand for stable, repeatable feature vectors.
Lightweight Model
Random Forest training keeps inference fast without expensive GPU requirements.
Real-time Feedback
On-screen predictions make it easy to validate each sign while you capture data.
From webcam to prediction
Collect, train, and deploy in a tight loop.
- 01 Capture labeled images for each sign.
- 02 Extract landmarks into `data.pickle`.
- 03 Train a classifier and save `model.p`.
- 04 Run live inference with on-screen predictions.
Deploy-ready, open source
Pair this frontend with your Python pipeline and ship a clean landing page for demos, class projects, or client work.
Quick start
python3 -m venv .venv
.venv/bin/python -m pip install -r requirements.txt
HAND_LANDMARKER_MODEL=./hand_landmarker.task .venv/bin/python inference_classifier.pyMeet the Developer
The mind behind SignSpeak
Rahim Shah
AI & Software Developer
Computer Science graduate focused on data science and applied machine learning, with experience building end-to-end ML systems across computer vision, NLP, and predictive analytics. Interested in developing reliable, real-world AI solutions with attention to data quality, performance, and practical impact.