app.py

import gradio as gr
from ultralyticsplus import YOLO, render_result
import cv2
import torch
import time

# Check GPU availability
print(f"CUDA available: {torch.cuda.is_available()}")
print(f"CUDA device count: {torch.cuda.device_count()}")
if torch.cuda.is_available():
    print(f"Current device: {torch.cuda.current_device()}")
    print(f"Device name: {torch.cuda.get_device_name(0)}")

# Load model with GPU acceleration
model = YOLO('foduucom/plant-leaf-detection-and-classification').to('cuda' if torch.cuda.is_available() else 'cpu')

# Model configuration
model.overrides['conf'] = 0.25
model.overrides['iou'] = 0.45
model.overrides['agnostic_nms'] = False
model.overrides['max_det'] = 1000

def detect_leaves(image):
    start_time = time.time()
    
    # Convert image and check processing time
    print("Converting image format...")
    img = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
    print(f"Conversion time: {time.time() - start_time:.2f}s")
    
    # Perform prediction with timing
    print("Starting prediction...")
    pred_start = time.time()
    results = model.predict(img, imgsz=640)  # Fixed inference size
    print(f"Prediction time: {time.time() - pred_start:.2f}s")
    
    # Process results
    num_leaves = len(results[0].boxes)
    print(f"Detected {num_leaves} leaves")
    
    # Render results
    print("Rendering results...")
    render_start = time.time()
    rendered_img = render_result(model=model, image=img, result=results[0])
    print(f"Rendering time: {time.time() - render_start:.2f}s")
    
    # Convert back to RGB
    return cv2.cvtColor(rendered_img, cv2.COLOR_BGR2RGB), num_leaves

# Create Gradio interface with queue
interface = gr.Interface(
    fn=detect_leaves,
    inputs=gr.Image(label="Upload Plant Image"),
    outputs=[
        gr.Image(label="Detected Leaves"),
        gr.Number(label="Number of Leaves Found")
    ],
    title="🍃 Plant Leaf Detection",
    allow_flagging="never"
)

if __name__ == "__main__":
    interface.launch(
        server_port=7860,
        share=False,
        enable_queue=True  # Essential for heavy computations
    )