BanglaTextAI

BanglaTextAI - Breaking Language Barriers with Emotional Intelligence

Inspiration

In our increasingly connected world, language barriers still prevent meaningful communication between English and Bengali speakers. We wanted to create a solution that not only translates text but also understands the emotional context of face-to-face interactions.

What it does:

BanglaTextAI is a dual-purpose application that combines:

1. Real-time bidirectional English-Bengali translation

2. Live facial emotion detection using machine learning

How we built it:

- Frontend: React.js with an intuitive, animated UI

- Backend: Flask server handling translation and ML inference

- ML Model: Custom CNN trained on FER2013 dataset (80.43% validation accuracy)

- Translation: Google Translate API integration

- Face Detection: OpenCV with Haar Cascade Classifier

Technical Features

- Bidirectional text translation

- Real-time emotion detection through webcam

- Interactive UI with smooth animations

- Robust error handling and fallback mechanisms

Challenges we overcame

- Implementing reliable face detection in various lighting conditions

- Ensuring accurate Bengali character rendering

- Optimizing ML model for real-time performance

- Creating seamless integration between React and Flask

Accomplishments

- Achieved 80.43% accuracy in emotion detection

- Created a responsive and intuitive user interface

- Successfully integrated multiple complex technologies

- Built a practical solution for real-world communication challenges

What we learned

- ML model deployment in production

- Real-time video processing

- Cross-language text processing

- Full-stack application architecture

What's next

- Offline translation support

- Mobile app development

- Multi-language expansion

- Enhanced emotion detection accuracy

- Real-time voice input

Built With

- React.js

- Flask

- TensorFlow

- OpenCV

- Google Translate API

Team: Script Kiddos