| title: Food Classifier with Model Comparison | |
| emoji: 🍔 | |
| colorFrom: green | |
| colorTo: blue | |
| sdk: gradio | |
| sdk_version: 4.19.2 | |
| app_file: app.py | |
| pinned: false | |
| # 🍔 Food Classifier: Accuracy vs. Speed | |
| This Gradio demo allows you to classify food images using two different transformer-based models and visually compare their performance. | |
| ## How to Use | |
| 1. **Upload an Image**: Drag and drop a food image or click to upload one. You can also use one of the examples below. | |
| 2. **Choose a Model**: Select either the ViT or Swin model from the dropdown. | |
| 3. **Click Classify**: The model will predict the food item. | |
| ## The Comparison Feature | |
| The key feature of this demo is the **performance comparison chart**: | |
| - **Benchmark Accuracy**: This chart shows the reported accuracy of each model on the Food101 test set. The Swin model is generally more accurate. | |
| - **Inference Time**: This chart shows the *actual time* it took for the selected model to process *your* uploaded image. You can see the speed trade-off firsthand. The ViT model is often faster. | |
| This allows you to understand the classic machine learning trade-off between a model's accuracy and its computational cost (speed). | |