Sign Language Detection using YOLOv5 v1.0.0
Features
- Real-time sign language detection via webcam
- Trained YOLOv5 model with bounding box localization
- Custom dataset training pipeline in Jupyter notebook
- Pre-trained weights (best.pt) included for instant inference
- Part of a comparative study with ANN and CNN approaches
Performance
- Real-time inference (30+ FPS on GPU)
- Bounding boxes with class labels and confidence scores
- Multi-sign detection in a single frame
Tech Stack
- Python 3.7+
- PyTorch
- YOLOv5 (Ultralytics)
- OpenCV
- Jupyter Notebook
Project Context
Final-year B.Tech 7th semester project comparing YOLOv5 object detection against traditional ANN and CNN classification approaches for sign language recognition.
Companion repository (ANN/CNN): https://github.com/zishnusarker/Sign-language-Detection-Using-ANN-CNN