SignLanguageRecognition

Live classification of sign language into words via webcam.

Installation

Install dependencies

pip install -r requirements.txt

Clone yolov5 into models/

git clone git@github.com:ultralytics/yolov5.git models/yolov5

Run

• Esc or q to close the window

python livedetection.py

Tested with words: friend, sister

• It opens a side-by-side window
  • Left: Live-Webcam with ROI, Right: Skeleton-Flow

Functionality

• Dataset for train and test:

• Mediapipe for skeleton

  • Pose
  • Hands

• OpenCV for image processing

• YOLOv5 for object detection (classification)

  • Model: YOLOv5s
  • trained model with YOLOv5 with command (with best results):

  python train.py --img 512 --batch 16 --epochs 100 --data models/yolov5_gesture/dataset.yaml --weights models/yolov5_gesture/yolov5s.pt