app.py

import gradio as gr
from ultralyticsplus import YOLO, render_result
import numpy as np
import time
import torch

# System Configuration
print("\n" + "="*40)
print(f"PyTorch: {torch.__version__}")
print(f"CUDA Available: {torch.cuda.is_available()}")
print("="*40 + "\n")

# Load model with optimized parameters for leaf counting
model = YOLO('foduucom/plant-leaf-detection-and-classification')

# Custom configuration for leaf counting
model.overrides.update({
    'conf': 0.15,  # Lower confidence threshold for better recall
    'iou': 0.25,    # Lower IoU threshold for overlapping leaves
    'imgsz': 1280,  # Higher resolution for small leaves
    'agnostic_nms': False,
    'max_det': 300,  # Higher maximum detections
    'device': 'cuda' if torch.cuda.is_available() else 'cpu',
    'classes': None,  # Detect all classes (leaves only in this model)
    'half': torch.cuda.is_available()
})

def count_leaves(image):
    try:
        start_time = time.time()
        
        # Preprocessing - enhance contrast
        image = np.array(image)
        lab = cv2.cvtColor(image, cv2.COLOR_RGB2LAB)
        l, a, b = cv2.split(lab)
        clahe = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8,8))
        cl = clahe.apply(l)
        limg = cv2.merge((cl,a,b))
        enhanced_img = cv2.cvtColor(limg, cv2.COLOR_LAB2RGB)
        
        # Prediction with overlap handling
        results = model.predict(
            source=enhanced_img,
            augment=True,  # Test time augmentation
            verbose=False,
            agnostic_nms=False,
            overlap_mask=False
        )
        
        # Post-processing for overlapping leaves
        boxes = results[0].boxes
        valid_boxes = []
        
        # Filter small detections and merge overlapping
        for box in boxes:
            x1, y1, x2, y2 = box.xyxy[0].cpu().numpy()
            w = x2 - x1
            h = y2 - y1
            
            # Filter too small boxes (adjust based on your leaf sizes)
            if w > 20 and h > 20:
                valid_boxes.append(box)
        
        # Improved NMS for overlapping leaves
        from utils.nms import non_max_suppression
        final_boxes = non_max_suppression(
            torch.stack([b.xywh[0] for b in valid_boxes]),
            conf_thres=0.1,
            iou_thres=0.15,
            multi_label=False
        )
        
        num_leaves = len(final_boxes)
        
        # Visual validation
        debug_img = enhanced_img.copy()
        for box in final_boxes:
            x1, y1, x2, y2 = map(int, box[:4])
            cv2.rectangle(debug_img, (x1, y1), (x2, y2), (0,255,0), 2)
        
        print(f"Processing time: {time.time()-start_time:.2f}s")
        return debug_img, num_leaves
    
    except Exception as e:
        print(f"Error: {str(e)}")
        return image, 0

# Gradio interface with visualization
interface = gr.Interface(
    fn=count_leaves,
    inputs=gr.Image(label="Input Image"),
    outputs=[
        gr.Image(label="Detection Visualization"),
        gr.Number(label="Estimated Leaf Count")
    ],
    title="🍃 Advanced Leaf Counter",
    description="Specialized for overlapping leaves and dense foliage",
    examples=[
        ["sample_leaf1.jpg"],
        ["sample_leaf2.jpg"]
    ]
)

if __name__ == "__main__":
    interface.launch(
        server_port=7860,
        share=False
    )