VisionGuard AI – Real-Time Phone Usage Detection

Inspiration

I wanted to build a project that solves a real-world problem using artificial intelligence and computer vision. Many students and employees lose focus because of constant phone usage during study or work sessions. I was inspired to create a system that can detect phone usage in real time and instantly notify the user.

I also wanted to improve my skills in AI, object detection, dataset creation, and real-time video processing.

What it does

VisionGuard AI detects objects and identifies when someone is using a phone through a live camera feed. The system processes frames in real time using a trained YOLOv8 model and computer vision techniques.

The project can:

Detect phones and faces in real time
Monitor live webcam activity
Notify users when phone usage is detected
Use custom datasets for better accuracy
Process video streams with low latency

How I built it

I built the project using Python and computer vision libraries. First, I collected and prepared datasets for object detection and face recognition. Then I trained and tested the YOLOv8 model for phone detection.

The application uses:

YOLOv8 for object detection
OpenCV for video processing
Dlib facial landmark detection
Python for the backend logic

I also created scripts for:

dataset generation
testing detections
validating labels
training workflows

Challenges I ran into

One of the biggest challenges was improving real-time detection performance while keeping the system responsive. Training datasets and tuning detection accuracy also took significant time.

Another challenge was organizing datasets and handling false detections in different lighting conditions and camera angles.