Skip to content

Latest commit

 

History

History
34 lines (28 loc) · 961 Bytes

README.md

File metadata and controls

34 lines (28 loc) · 961 Bytes

ScamperWang

The internet of spiders

Deep Learning classifier built with fast.ai and PyTorch to recognise 8 species of spider

  • House spider (Tegenaria domestica)
  • Giant House spider (Eratigena atrica)
  • Zebra Spider (Salticus scenicus)
  • Sector Spider (Zygiella x-notata)
  • Cellar Spider (Pholcidae)
  • Cucumber Spider (Araniella cucurbitina)
  • Cross Orb Weaver (Araneus diadematus)
  • False Widow (Steatoda nobilis)

The server is a Starlette rest server written in python. The architecture is Resnet50 and was trained using this Jupyter notebook on a Colab GPU: https://github.com/peetdenny/FastAI/blob/master/ScamperWang.ipynb

Prerequisites

You should have Python 3 installed with pip. And then run

pip install uvicorn
pip install fastai
pip install aiohttp
pip install aiofiles
pip install python-multipart

Running the server

python server.py serve

Then point your brower at http://localhost:8008 and upload a photo of a spider!