perceptionpro 0.1.3

PerceptionPro is a package for computer vision tasks such as head pose estimation, eye tracking, and object detection.

PerceptionPro

PerceptionPro is a Python package designed for real-time monitoring and alert systems. It provides modular components for head pose estimation, eye tracking, and object detection, all integrated into a cohesive alert system.

Features

• Head Pose Estimation: Tracks the orientation of the user's head in real-time, providing insights into the direction of attention.

• Eye Tracking: Detects and analyzes eye movements and gaze direction, ensuring effective focus monitoring.

• Object Detection: Identifies objects in the environment, supporting compliance and situational awareness.

• Alert System: Integrated mechanism to trigger alerts based on configurable thresholds for head pose, eye tracking, and object detection.

Installation

pip install perceptionpro

Components

1. Head Pose Estimation:

• Module: HeadPoseEstimator

• Uses facial landmarks to determine the orientation of the user's head (e.g., Left, Right, Up, Down, Center).

2. Eye Tracking:

• Module: EyeTracker

• Tracks eye movements and gaze direction for applications like focus monitoring and attention analysis.

3. Object Detection:

• Module: ObjectDetector

• Leverages the YOLO model to identify and count objects such as person, book, and cell phone in real-time.

4. Alert System:

• Threshold-based system to monitor head pose, eye tracking, and detected objects, triggering alerts when defined criteria are met.

Usage

import cv2
import time
import keyboard 
from perceptionpro.core import PerceptionInit

def main():
    """
    Main function to initialize the camera, process video frames,
    and handle user input to quit.
    """
    try:
        # Initialize the camera
        camera = cv2.VideoCapture(0)

        if not camera.isOpened():
            print("Error: Camera could not be opened.")
            return
        else:
            print("Camera opened successfully.")

        # Initialize the PerceptionInit with speech enabled
        vision = PerceptionInit(camera, model_path="https://github.com/UmarBalak/ProctorVision/raw/refs/heads/main/yolov8s.pt")

        # Initialize variables
        violation_count = 0
        prev_time = time.time()

        while True:
            # Capture frame-by-frame
            ret, frame = camera.read()

            if not ret:
                print("Failed to capture video. Check your camera connection.")
                break

            # Calculate frame rate
            current_time = time.time()
            elapsed_time = current_time - prev_time
            fps = 1 / elapsed_time if elapsed_time > 0 else 0
            prev_time = current_time

            # Process frame metrics
            result, eye_d, head_d, fps, obj_d, alert_msg = vision.track()
            print("Procesed")

            if not result:
                violation_count += 1
                print(f"Warning: {violation_count} - {alert_msg}")

                if violation_count == 4:
                    print("The exam has been terminated.")
                    break
            else:
                pass
                # Print real-time metrics to console
                print(f"FPS: {fps:.2f}")
                print(f"Eye Direction: {eye_d}")
                print(f"Head Direction: {head_d}")
                print(f"Background: {'Ok' if obj_d else 'Object detected'}")

            # Check if 'q' is pressed to exit the loop
            if keyboard.is_pressed('q'):
                print("User requested to stop the process.")
                break

    except Exception as e:
        print(f"An unexpected error occurred: {e}")

    finally:
        # Always release the camera and close windows
        if 'camera' in locals() and camera.isOpened():
            camera.release()
        print("Camera released. Process complete.")

if __name__ == "__main__":
    main()