This project is part of our research into building an intelligent prosthetic robotic arm that can understand human hand gestures through computer vision and AI.
We are collecting and preprocessing image data of different hand gestures to train machine learning models. The ultimate goal is to connect these models to a prosthetic arm, enabling it to perform natural movements based on user intent.
- data_creation.py – Script for recording hand gesture images using a webcam.
- merge_datasets.py – Script for merging multiple gesture datasets into one.
- gesture_recognition.ipynb – Jupyter notebook for testing, training, and experimenting with models.
- Investigate how AI can interpret hand gestures reliably.
- Build datasets that can generalize to different users and conditions.
- Lay the foundation for gesture-controlled prosthetics, where computer vision acts as an interface between human movement and robotic motion.
- Train CNN or transformer-based models on the gesture dataset.
- Integrate trained models with Arduino Mega + servos for real-time control.
- Experiment with multimodal input (e.g., EMG signals + vision).