app.py

import gradio as gr
import cv2
import torch
import numpy as np

# Load the YOLOv5 model
model = torch.hub.load('ultralytics/yolov5', 'yolov5s', pretrained=True)

# Function to run inference on an image
def run_inference(image):
    # Convert the image from PIL format to a format compatible with OpenCV
    image = np.array(image)

    # Run YOLOv5 inference
    results = model(image)

    # Convert the annotated image from BGR to RGB for display
    annotated_image = results.render()[0]
    annotated_image = cv2.cvtColor(annotated_image, cv2.COLOR_BGR2RGB)

    return annotated_image

# Function to generate a summary for the detected objects
def generate_summary(image):
    results = model(image)
    detected_objects = results.pandas().xyxy[0]
    summary = "Detected objects:\n\n"
    for idx, obj in detected_objects.iterrows():
        summary += f"- {obj['name']} with confidence {obj['confidence']:.2f}\n"
    return summary

# Create the Gradio interface with improved UI
with gr.Blocks(css="""
    body {
        font-family: 'Poppins', sans-serif;
        background-color: #2B3D41;
        color: #F9B9D2;
    }
    header {
        background-color: #83A0A0;
        padding: 20px;
        text-align: center;
        border-radius: 10px;
        color: white;
        box-shadow: 0 4px 10px rgba(0, 0, 0, 0.3);
    }
    footer {
        background-color: #4C5F6B;
        padding: 10px;
        text-align: center;
        border-radius: 10px;
        color: white;
        margin-top: 20px;
        box-shadow: 0 4px 10px rgba(0, 0, 0, 0.3);
    }
    .btn-primary {
        background-color: #BCA0BC;
        color: #2B3D41;
        padding: 10px 20px;
        border-radius: 5px;
        font-weight: bold;
        border: none;
        cursor: pointer;
        transition: all 0.3s;
    }
    .btn-primary:hover {
        background-color: #F9B9D2;
        color: #2B3D41;
    }
    .gr-box {
        background-color: #4C5F6B;
        border-radius: 10px;
        padding: 20px;
        color: #F9B9D2;
        box-shadow: 0 4px 10px rgba(0, 0, 0, 0.3);
    }
    .gr-input {
        background-color: #BCA0BC;
        border-radius: 5px;
        border: none;
        padding: 10px;
        color: #2B3D41;
    }
""") as demo:
    with gr.Row():
        gr.Markdown("<h1 style='text-align:center; color:#F9B9D2;'>✨ InsightVision: Detect, Analyze, Summarize ✨</h1>")
    
    with gr.Row():
        with gr.Column(scale=2):
            image_input = gr.Image(label="Upload Image", type="pil", elem_classes="gr-input")
            with gr.Row():
                detect_button = gr.Button("Run Detection", elem_classes="btn-primary")
        with gr.Column(scale=3):
            annotated_image_output = gr.Image(label="Detected Image", type="pil", elem_classes="gr-box")
            summary_output = gr.Textbox(label="Detection Summary", lines=10, interactive=False, elem_classes="gr-box")
    
    # Actions for buttons
    detect_button.click(
        fn=lambda image: (run_inference(image), generate_summary(np.array(image))),
        inputs=[image_input],
        outputs=[annotated_image_output, summary_output]
    )

    gr.Markdown("<footer>Made with ❤️ using Gradio and YOLOv5 | © 2024 InsightVision</footer>")

# Launch the interface
demo.launch()