ASL Communication Assistant — Computer Vision Prototype

ASL Communication Assistant

Personal computer-vision and machine-learning prototype developed by Ali Raza. This is a local learning/research project, not a paid client deployment.

The application uses Python, MediaPipe and OpenCV to capture webcam hand landmarks, classify a defined vocabulary of signs and assemble recognized tokens into editable text. It includes data collection, model-training scripts, temporal smoothing and a desktop application workflow.

The cover is a project overview graphic, not an application screenshot. The implementation and saved evaluation artifacts were inspected locally. No universal ASL translation capability or production accuracy is claimed: recognition depends on the vocabulary, training data, signer and environment.

The optional WLASL-derived word-model experiment is an academic/non-commercial research component. This portfolio demonstrates the software workflow without distributing datasets, model weights or private project files.