|
--- |
|
title: Bean Plant Health ViT Classifier |
|
emoji: 🌱📸🩺 |
|
colorFrom: green |
|
colorTo: green |
|
sdk: gradio |
|
sdk_version: 4.41.0 |
|
app_file: app.py |
|
pinned: false |
|
license: afl-3.0 |
|
--- |
|
|
|
# Bean Plant Health Predictor |
|
|
|
This application is designed to help farmers quickly identify the health of bean plants by analyzing images of their leaves. The app uses a Vision Transformer (ViT) model to classify images into three categories: **angular_leaf_spot**, **bean_rust**, and **healthy**. This tool can be deployed on a drone for real-time monitoring of crops, enabling timely treatment of diseased plants. |
|
|
|
## Use Case |
|
|
|
Farmers need to monitor the health of their bean plants regularly to prevent the spread of diseases. This app provides a machine learning-based solution to automate the identification of plant diseases, which can be particularly useful when integrated with drone technology. |
|
|
|
## Features |
|
|
|
- **Image Classification**: Upload an image of a bean leaf, and the app will classify it into one of the following categories: |
|
- Angular Leaf Spot |
|
- Bean Rust |
|
- Healthy |
|
|
|
## Model Details |
|
|
|
- **Model Used**: [Vision Transformer (ViT) - base-sized model](https://huggingface.co/google/vit-base-patch16-224) fine-tuned on the [Beans dataset](https://huggingface.co/datasets/beans). |
|
- **Image Processor**: The app uses the `ViTImageProcessor` for preparing images before classification. |
|
- **Labels**: The possible outcomes are `angular_leaf_spot`, `bean_rust`, and `healthy`. |
|
|
|
## How to Use |
|
|
|
1. **Upload an Image**: Click on the image input field and upload a photo of a bean leaf. |
|
2. **Get Results**: The app will classify the image and display the probabilities for each category. |
|
3. **Interpret the Results**: The app shows the confidence levels for each label, helping farmers identify whether the plant is healthy or requires treatment. |
|
|
|
## Technology Stack |
|
|
|
- **Gradio**: Used to create the user interface. |
|
- **PyTorch**: Utilized for running the model inference. |
|
- **Hugging Face Transformers**: Provides the pre-trained Vision Transformer model. |
|
|
