| from fastai.vision.all import * | |
| import gradio as gr | |
| from PIL import Image | |
| import numpy as np | |
| # Load the model | |
| learn = load_learner('export.pkl') | |
| categories = learn.dls.vocab | |
| # Create a mapping from original label → display label | |
| pretty_labels = {cat: cat.replace('_', ' ').title() for cat in categories} | |
| # Display-friendly veggie list | |
| veggie_list = ', '.join(pretty_labels[cat] for cat in categories) | |
| description = f"Upload a vegetable image to classify it.\n\n The model can recognize: {veggie_list}." | |
| # Prediction function | |
| def classify_image(img): | |
| if isinstance(img, str): # If it's a filepath | |
| img = Image.open(img) | |
| elif isinstance(img, np.ndarray): # If it's a NumPy array | |
| img = Image.fromarray(img) | |
| # If it's already a PIL Image, no conversion needed | |
| pred, idx, probs = learn.predict(img) | |
| return { | |
| pretty_labels[c]: float(probs[i]) | |
| for i, c in enumerate(categories) | |
| } | |
| # Gradio UI | |
| image = gr.Image(type="pil", label="Upload a vegetable image") | |
| label = gr.Label(label="Prediction") | |
| intf = gr.Interface( | |
| fn=classify_image, | |
| inputs=image, | |
| outputs=label, | |
| title="Vegetable Classifier", | |
| description=description | |
| ) | |
| intf.launch() | |