---
title: Traffice Light
emoji: 🔥
colorFrom: indigo
colorTo: indigo
sdk: gradio
sdk_version: 5.47.2
app_file: app.py
pinned: false
---

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

# 🚦 Traffic Sign Identifier

An interactive image classification app for identifying traffic signs using a pretrained **AutoGluon MultiModalPredictor** model hosted on Hugging Face.  
The app provides **predictions with confidence scores**, a clean **Gradio interface**, and user-friendly visualizations.

---

## ✨ Features

- **Traffic Sign Classification**  
  Upload or capture an image of a traffic sign and receive predictions.

- **Model Integration**  
  Uses `cassieli226/sign-identification-automl` trained with AutoGluon, deployed via Hugging Face Hub.

- **Configurable Inference**  
  - Resize size: adjust input resolution (64–512 px).  
  - Top-k predictions: see multiple likely classes.  
  - Probability threshold: filter low-confidence results.

- **Interactive Interface**  
  Built with [Gradio](https://gradio.app/) for intuitive user experience, including drag-and-drop upload and webcam capture.

- **Styled Output**  
  Results are presented with clear visuals: top prediction, confidence %, and a ranked list of alternatives.

---

## 📂 Project Structure

- **`app.py`** — main application with Gradio Blocks interface.  
- **`autogluon_predictor_dir.zip`** — packaged AutoGluon model checkpoint (downloaded from Hugging Face Hub).  
- **`requirements.txt`** — dependencies for running locally or in a Hugging Face Space.  
- **`README.md`** — this documentation.  

---

## 🚀 Running the App

### 1. Local Setup
Clone the repository and install dependencies:
```bash
pip install -r requirements.txt
```
## 📊 Example Usage

1. Upload an image of a traffic sign (or use webcam capture).  
2. Adjust resize size, top-k predictions, or probability threshold if desired.  
3. Click **Predict**.  

You’ll see:  
- Original image preview.  
- Top predicted sign + confidence score.  
- Ranked list of additional predictions with probabilities.  

---

## 📚 Citations & References

- **AutoGluon**: Erickson et al., *AutoGluon: AutoML Toolkit for Deep Learning*, [GitHub](https://github.com/autogluon/autogluon).  
- **Gradio**: Abid et al., *Gradio: Hassle-Free Sharing and Testing of ML Models*, [gradio.app](https://gradio.app).  
- **Hugging Face Hub** for model hosting.  

---

## 📜 License

This project is distributed under the **MIT License**. See [LICENSE](LICENSE) for details.  

---

## 🙌 Acknowledgments

- **Model** trained by `cassieli226` and shared via Hugging Face Hub.  
- **App** adapted and deployed by `maryzhang`.  
- Special thanks to classmates and instructors for feedback.  

---

## 🤖 AI Usage

This project made use of **ChatGPT (OpenAI)** during development to:  
- Refactor and debug the Gradio Blocks interface.  
- Improve prediction display styling with HTML/CSS.  
- Draft and polish this README for clarity and completeness.