app.py

import cv2
import numpy as np
import torch
import gradio as gr
from ultralytics import YOLO

# Load YOLOv12x model
MODEL_PATH = "yolov12x.pt"  # Ensure the model is uploaded to the Hugging Face Space
model = YOLO(MODEL_PATH)

# COCO dataset class ID for trucks
TRUCK_CLASS_ID = 7  # "truck"

def count_trucks(video_path):
    cap = cv2.VideoCapture(video_path)
    if not cap.isOpened():
        return "Error: Unable to open video file."

    frame_count = 0
    truck_counts = []
    frame_skip = 5  # Process every 5th frame for efficiency

    while True:
        ret, frame = cap.read()
        if not ret:
            break  # End of video

        frame_count += 1
        if frame_count % frame_skip != 0:
            continue  # Skip frames to improve efficiency

        # Run YOLOv12x inference
        results = model(frame, verbose=False)

        truck_count = 0
        for result in results:
            for box in result.boxes:
                class_id = int(box.cls.item())  # Get class ID
                confidence = float(box.conf.item())  # Get confidence score

                # Count only trucks
                if class_id == TRUCK_CLASS_ID and confidence > 0.5:
                    truck_count += 1

        truck_counts.append(truck_count)

    cap.release()

    return {
        "Trucks in a Frame": int(np.max(truck_counts)) if truck_counts else 0
    }

# Gradio UI function
def analyze_video(video_file):
    result = count_trucks(video_file)
    return "\n".join([f"{key}: {value}" for key, value in result.items()])

# Define Gradio interface
iface = gr.Interface(
    fn=analyze_video,
    inputs=gr.Video(label="Upload Video"),
    outputs=gr.Textbox(label="Truck Count"),
    title="YOLOv12x Truck Counter",
    description="Upload a video to count trucks using YOLOv12x."
)

# Launch the Gradio app
if __name__ == "__main__":
    iface.launch()