README.md

---
title: Stop Sign Classifier — Original vs. Preprocessed
emoji: 🛑
colorFrom: red
colorTo: blue
sdk: gradio
sdk_version: "4.44.0"
app_file: app.py
pinned: false
---

# Stop Sign? — Image Classifier (Classmate Model)

This Space demonstrates a **classmate’s AutoGluon MultiModal** image model that predicts whether a **stop sign** is present in an image.

- **Model (classmate’s):** [samder03/2025-24679-image-autogluon-predictor](https://huggingface.co/samder03/2025-24679-image-autogluon-predictor)
- **Dataset (source/docs):** https://huggingface.co/datasets/ecopus/sign_identification

## What you can do here
- Upload a PNG/JPG (or use your webcam).
- See the **original** image next to the **preprocessed** image the model actually sees (resize + optional center-crop).
- Adjust **inference parameters**: resize (shorter side), center-crop on/off, crop size, and Top-K classes.
- View ranked **Top-K** class probabilities.

## Validation & Error Handling
- File type is constrained by the Image widget (PNG/JPG; webcam allowed).
- Very large images are rejected with a friendly message (pixel-count check).

## Notes
Educational demo; not for safety-critical use. Outputs depend on dataset/model behavior. The interface and upload code was created with the assistance of GenAI