app.py

import os
import cv2
import numpy as np
import torch
from ultralytics import YOLO
from sort import Sort
import gradio as gr

# Load YOLOv12x model
MODEL_PATH = "yolov12x.pt"
model = YOLO(MODEL_PATH)

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

# Initialize SORT tracker
tracker = Sort(max_age=20, min_hits=3, iou_threshold=0.3)  # Improved tracking stability

# Minimum confidence threshold for detection
CONFIDENCE_THRESHOLD = 0.4  # Adjusted to capture more trucks

# Distance threshold to avoid duplicate counts
DISTANCE_THRESHOLD = 50

# Dictionary to define keyword-based time intervals
TIME_INTERVALS = {
    "one": 1, "two": 2, "three": 3, "four": 4, "five": 5,
    "six": 6, "seven": 7, "eight": 8, "nine": 9, "ten": 10, "eleven": 11
}

def determine_time_interval(video_filename):
    """ Determines frame skip interval based on keywords in the filename. """
    for keyword, interval in TIME_INTERVALS.items():
        if keyword in video_filename:
            return interval
    return 5  # Default interval

def count_unique_trucks(video_path):
    """ Counts unique trucks in a video using YOLOv12x and SORT tracking. """
    cap = cv2.VideoCapture(video_path)
    if not cap.isOpened():
        return {"Error": "Unable to open video file."}

    unique_truck_ids = set()
    truck_history = {}

    # Get FPS of the video
    fps = int(cap.get(cv2.CAP_PROP_FPS))

    # Extract filename from the path and convert to lowercase
    video_filename = os.path.basename(video_path).lower()

    # Determine the dynamic time interval based on filename keywords
    time_interval = determine_time_interval(video_filename)

    # Get total frames in the video
    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))

    # Dynamically adjust frame skipping based on FPS and movement density
    frame_skip = max(1, min(fps * time_interval // 2, total_frames // 10))

    frame_count = 0

    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 based on interval

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

        detections = []
        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

                # Track only trucks
                if class_id == TRUCK_CLASS_ID and confidence > CONFIDENCE_THRESHOLD:
                    x1, y1, x2, y2 = map(int, box.xyxy[0])  # Get bounding box
                    detections.append([x1, y1, x2, y2, confidence])

        # Convert detections to numpy array for SORT
        detections = np.array(detections) if len(detections) > 0 else np.empty((0, 5))

        # Update SORT tracker
        tracked_objects = tracker.update(detections)

        # Track movement history to avoid duplicate counts
        for obj in tracked_objects:
            truck_id = int(obj[4])  # Unique ID assigned by SORT
            x1, y1, x2, y2 = obj[:4]  # Get bounding box coordinates

            truck_center = (x1 + x2) / 2, (y1 + y2) / 2  # Calculate truck center

            # Entry-exit zone logic (e.g., bottom 20% of the frame)
            frame_height, frame_width = frame.shape[:2]
            entry_line = frame_height * 0.8  # Bottom 20% of the frame
            exit_line = frame_height * 0.2  # Top 20% of the frame

            if truck_id not in truck_history:
                # New truck detected
                truck_history[truck_id] = {
                    "position": truck_center,
                    "crossed_entry": truck_center[1] > entry_line,
                    "crossed_exit": False
                }
                continue

            # If the truck crosses from entry to exit, count it
            if truck_history[truck_id]["crossed_entry"] and truck_center[1] < exit_line:
                truck_history[truck_id]["crossed_exit"] = True
                unique_truck_ids.add(truck_id)

    cap.release()
    return {"Total Unique Trucks": len(unique_truck_ids)}

# Gradio UI function
def analyze_video(video_file):
    result = count_unique_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="Analysis Result"),
    title="YOLOv12x Unique Truck Counter",
    description="Upload a video to count unique trucks using YOLOv12x and SORT tracking."
)

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