How prem-buraga/automated-mobile-phone-usage-detection-and-mailing-system-using-deep-learning-techniques works
Main Components
• Camera Feed / Video Input – Represents the source of real-time images or frames.
• Detection Module – Contains two sub-modules:
• Object Detection using YOLOv8 with the pre-trained model (best.pt) to identify mobile phone usage.
• Face Detection using Haar Cascade (haarcascade_frontalface_default.xml) for extracting face images.
• Data Storage & Logging – Uses files such as Details.xlsx to store user and incident details and folders for storing captured images (e.g., “captured_faces/” and “captured_frames/” as mentioned in README, though not shown explicitly in the file tree).
• Notification System – The automated mailing module that sends out alerts with incident details (including images, timestamps, etc.).
• Control Script – The Mobile_usage_detection.py, which orchestrates the flow between modules.
• Configuration & Development Tools – The .vscode folder and potential configuration files that may help with development, debugging, or deployment.
Relationships and Interactions
• The Camera Feed continuously supplies video frames to the Control Script.
• The Control Script passes these frames to the Object Detection module (YOLOv8) to detect mobile phone use.
• For detected incidents, the same frame is processed by the Face Detection module to capture and log face images.
• Once an incident is confirmed, the system records details (timestamp, captured images) and cross-references details from Details.xlsx as necessary.
• The Notification System then automatically sends emails containing incident logs, images, and other relevant information.
Architectural Patterns & Important Design Principles
• Modularity: Separate components (detection modules, notification system, data storage) allow for independent updates or troubleshooting.
• Real-Time Processing: Emphasize components handling streaming data from the camera.
• Integration of Machine Learning: The YOLOv8 (for object detection) and Haar Cascade (for face detection) modules highlight the use of pre-trained models integrated into an application pipeline.
• External Dependency Integration: Automated email notifications and using an Excel sheet for details suggest that integration with external systems is important.
Technologies, Frameworks, and Libraries
• Python for scripting, with libraries such as OpenCV, numpy, pandas, face_recognition, and ultralytics.
• YOLOv8 (via ultralytics) for object detection.
• Haar Cascade (via OpenCV) for face recognition.
• Excel file processing using openpyxl or pandas.