ASL YOLOv11 — Real-Time Sign Language Detection interface and evaluation output
Figure 01 — ASL RecognitionOpen full resolution ↗

notesarchitecture & scope

A real-time computer vision pipeline that localizes and classifies the 26-letter American Sign Language (ASL) alphabet from live video streams. Built around YOLOv11 object detection and custom dataset curation in Roboflow.

Key Capabilities

  • Low-latency hand localization and bounding-box inference via YOLOv11
  • 26-class alphabet gesture classification achieving 95%+ validation precision
  • Live webcam inference pipeline built with OpenCV
  • Custom raw-image annotation, bounding-box QA, and augmentation pipeline on Roboflow
  • Confusion-matrix, precision-recall, and F1 evaluation diagnostics across lighting conditions

Designed as a practical exploration into accessible human-computer interaction and edge-ready vision models.