app.py

import gradio as gr
from fastai.vision.all import *
from fastai.learner import load_learner
from pathlib import Path
import pandas as pd
import os

"""
Warning Lamp Detector using FastAI
This application allows users to upload images of warning lamps and get classification results.
"""

def get_labels(fname):
    """
    Function required by the model to process labels
    Args:
        fname: Path to the image file
    Returns:
        list: List of active labels
    """
    # Since we're only doing inference, we can return an empty list
    # This function is only needed because the model was saved with it
    return []

# Load the FastAI model
try:
    model_path = Path("WarningLampClassifier.pkl")
    learn_inf = load_learner(model_path)
    print("Model loaded successfully")
except Exception as e:
    print(f"Error loading model: {e}")
    raise

def detect_warning_lamp(image, history: list[tuple[str, str]], system_message):
    """
    Process the uploaded image and return detection results using FastAI model
    Args:
        image: PIL Image from Gradio
        history: Chat history
        system_message: System prompt
    Returns:
        Updated chat history with prediction results
    """
    try:
        # Convert PIL image to FastAI compatible format
        img = PILImage(image)
        
        # Get model prediction
        pred_class, pred_idx, probs = learn_inf.predict(img)
        
        # Convert tensor outputs to Python types
        pred_class = str(pred_class)  # Convert class name to string
        pred_idx = int(pred_idx)  # Convert index to integer
        probs = [float(p) for p in probs]  # Convert probabilities to float list
        
        # Format the prediction results
        confidence = probs[pred_idx]  # Get confidence for predicted class
        response = f"Detected Warning Lamp: {pred_class}\nConfidence: {confidence:.2%}"
        
        # Add probabilities for all classes
        response += "\n\nProbabilities for all classes:"
        for cls, prob in zip(learn_inf.dls.vocab, probs):
            response += f"\n- {cls}: {prob:.2%}"
            
        # Update chat history
        history.append((None, response))
        return history
    except Exception as e:
        error_msg = f"Error processing image: {str(e)}"
        history.append((None, error_msg))
        return history

# Create a custom interface with image upload
with gr.Blocks(title="Warning Lamp Detector", theme=gr.themes.Soft()) as demo:
    gr.Markdown("""
    # 🚨 Warning Lamp Detector
    Upload an image of a warning lamp to get its classification.
    
    ### Instructions:
    1. Upload a clear image of the warning lamp
    2. Wait for the analysis
    3. View the detailed classification results
    
    ### Supported Warning Lamps:
    """)
    
    # Display supported classes if available
    if 'learn_inf' in locals():
        gr.Markdown("\n".join([f"- {cls}" for cls in learn_inf.dls.vocab]))
    
    with gr.Row():
        with gr.Column(scale=1):
            image_input = gr.Image(
                label="Upload Warning Lamp Image",
                type="pil",
                sources="upload"
            )
            system_message = gr.Textbox(
                value="You are an expert in warning lamp classification. Analyze the image and provide detailed information about the type, color, and status of the warning lamp.",
                label="System Message",
                lines=3,
                visible=False  # Hide this since we're using direct model inference
            )
        
        with gr.Column(scale=1):
            chatbot = gr.Chatbot(
                [],
                elem_id="chatbot",
                bubble_full_width=False,
                avatar_images=(None, "🚨"),
                height=400
            )
    
    # Add a submit button
    submit_btn = gr.Button("Analyze Warning Lamp", variant="primary")
    submit_btn.click(
        detect_warning_lamp,
        inputs=[image_input, chatbot, system_message],
        outputs=chatbot
    )

if __name__ == "__main__":
    demo.launch()