Pathos – Real-time Face Recognition and Emotion Classification

About

Pathos is a real-time face recognition and emotion classification tool that detects faces on the screen using the DeepFace framework. It supports state-of-the-art facial recognition models such as:

• VGG-Face
• FaceNet
• OpenFace
• DeepID
• ArcFace
• Dlib
• SFace
• GhostFaceNet

The tool classifies detected facial expressions into the following emotion categories:

• Happy
• Sad
• Anger
• Fear
• Surprise
• Disgust
• Neutral

Features

• Face Recognition: Detects faces on-screen using advanced models
• Emotion Classification: Identifies 7 core emotions in real time
• Screen Capture: Uses Chrome APIs to analyze tab content
• Canvas Overlay: Visualizes detection on screen
• Chrome Extension UI: Interactive popup with real-time control
• Python Backend: Handles image decoding, emotion recognition, and response handling

Technology Stack

Frontend

• React.js
• JavaScript (Vanilla)
• HTML5 / CSS3
• Chrome Extension APIs
• Canvas API
• RequestAnimationFrame for smooth rendering

Backend

• Python
• Flask
• DeepFace
• OpenCV (cv2)
• NumPy

Chrome Extension Architecture

• Manifest V3
• Background service worker
• Content scripts
• Popup interface
• Message passing (runtime & tabs API)

Integration

• Image Transfer: Captured screen frames are encoded in Base64 and sent to Flask backend
• Real-time Classification: Backend returns emotion classification as JSON
• Asynchronous Data Flow: Uses async/await and event-driven updates for responsiveness