Isnarjr/ASL-Sign-Recognition
🖐 Recognize American Sign Language gestures in real-time using hand tracking and machine learning for accessible communication.
What's novel
🖐 Recognize American Sign Language gestures in real-time using hand tracking and machine learning for accessible communication.
Code Analysis
5 files read · 2 roundsA three-script ASL letter recognition pipeline that captures hand landmark images via MediaPipe, trains a small MLP on wrist-relative normalized 42-dim landmark vectors, and performs real-time webcam classification of A-Z signs.
Strengths
The wrist-relative, scale-normalized landmark feature engineering is a clean and effective approach that makes the model invariant to hand position and size, and the three-script separation (capture/train/predict) is well-organized for the task.
Weaknesses
No tests exist, the README is heavily padded with broken links and marketing copy that misrepresents the project as a downloadable application rather than a set of Python scripts, and there's a potential NameError bug in capture_dataset.py where 'crop' could be referenced before assignment.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
9
Contributors
2
Files
10
Active weeks
3
Repository
Language
Python
Stars
3
Forks
2
License
—