A TL-based vision model that recognize 10 specific medical instruments from images.
This model can classify 10 types of medical instruments:
- Scalpel, Thermometer, Forceps, Bone Saw, Reflex Hammer
- Ultrasound Probe, Defibrillator, Otoscope, ECG Machine, Glucometer
Dataset: Collected using DuckDuckGo, cleaned using FastAI’s ImageClassifierCleaner.
Training: Fine-tuned a DenseNet121 model, achieving ~90% accuracy. Find details in notebooks.
Deployment: Hosted on Hugging Face Spaces with Gradio for easy interaction. See deployment folder.
Web API: Integrated with GitHub Pages for real-time classification.
- Gradio App → Test the Model
- GitHub Pages API Integration → Live Web App
- FastAI + PyTorch for training
- Gradio + Hugging Face Spaces for deployment
- GitHub Pages for API integration
- Google Colab for development