This is a Next.js app that classifies 70 different dog breeds, trained with Efficient Net B0, achieved ~90% accuracy.
The model is trained using PyTorch, can be viewed in /pytorch-training
folder.
After training, the model is saved as ONNX format. The model is loaded and inferenced on edge using ONNX Runtime Web API, thus has a low edge inference time.
WebApp
- Next.js
- Tailwind
- ONNX Runtime Web API
- Edge Inference
Machine Learning Model
- PyTorch
- ResNet18 and EffNet B0
- Kaggle Dataset - here
First, run the development server:
npm run dev
# or
yarn dev
# or
pnpm dev
Open http://localhost:3000 with your browser to see the result.