README.md

metadata

title: Door and Window Detector
colorFrom: gray
colorTo: blue
sdk: streamlit
app_file: app.py
pinned: false

doors-and-windows

an application that serves inference to detect doors and windows from architecture / floor plans.

step 0: learn to read architecture plans to distinguish between doors and windows. (DONE)

1. annotate data and save the labels. (DONE)
2. train the model, explain the difference between yolov8n and yolov8s, show which one is better. (DONE)
3. api development: will be a little hard. (DONE)
4. full cleanup (DONE) + streamlit setup (DONE)
5. deployment onto hf spaces + docker. (DONE)

extras:

1. linting + formatting.
2. mention model stats, like inference time, precision, etc.
3. add visualization by integrating wandb / tensorboard (user choice)
4. smooth web app and create api contract.