Computer Vision

Real-time object detection and face recognition using webcam with advanced AI models powered by TensorFlow.js and Google MediaPipe.

Real-time Object Detection

Loading AI Model...

0
FPS
0
Objects
Idle
Status
50%
Confidence
0%50%100%

Real-time Face Detection

Loading AI Models...

Powered by Google MediaPipe

0
FPS
0
Faces
Idle
Status

This feature uses Google MediaPipe Face Detection running entirely in your browser. All processing happens locally - no video or images are sent to any server. Your privacy is fully protected.

About COCO-SSD Model

Model Details

COCO-SSD (Single Shot MultiBox Detection) is a pre-trained object detection model that can identify and locate 80 different object categories in real-time.

Object Classes
80
Framework
TensorFlow.js

Detectable Objects

personbicyclecarmotorcycleairplanebustraintruckboattraffic lightfire hydrantstop signparking meterbenchbirdcatdoghorsesheepcow+60 more

Features

  • Real-time detection on modern devices
  • Runs entirely in browser (no server required)
  • Bounding box visualization with confidence scores
  • Adjustable confidence threshold for filtering

Privacy

All processing happens locally in your browser. No video or images are sent to any server. Your camera feed is private and secure.

YOLO Object Detection

Server-Side

Upload an image to detect objects using YOLOv8 - more accurate detection powered by our backend.

This uses YOLOv8 nano model running on our backend server for more accurate detection. Images are processed securely and not stored.

Technology Stack

TensorFlow.js

Machine learning library that runs in the browser, enabling real-time inference without server calls.

TensorFlow.jsCOCO-SSDWebGL

WebRTC & Canvas

Browser APIs for camera access and real-time graphics rendering for bounding box visualization.

WebRTCCanvas APIMediaDevices

React Hooks

Custom hooks for managing camera state, model loading, and detection lifecycle with performance optimization.

React 19TypeScriptCustom Hooks

Performance

Real-time FPS monitoring, confidence threshold filtering, and optimized rendering for smooth detection.

~30 FPSClient-sideNo Server

YOLOv8 (Server-Side)

Advanced object detection using YOLOv8 on the backend for higher accuracy and more robust detection.

YOLOv8PythonFastAPI

Face Detection (MediaPipe)

Real-time face detection powered by Google's MediaPipe running entirely in the browser with high performance and accuracy.

MediaPipeGoogle AIClient-side