Tell us about your project — BoloBetter 🌐

BoloBetter is an AI-powered language translation project designed to make communication easier between English, Hindi, and Hinglish.

Users can enter text directly or upload an image to extract the text using OCR. The extracted text can then be used for translation, helping users understand text found in screenshots, documents, posters, and other images.

The project uses the Gemma language model and explores LoRA fine-tuning to improve natural, conversational translations. It also includes experiments to evaluate translation quality and response speed.

Key Features

• 💬 Text Translation: Translate text between supported languages.
• 🖼️ Image-to-Text Extraction: Upload an image and extract its text using OCR.
• 🌐 Multilingual Support: Designed for English, Hindi, and Hinglish workflows.
• 🧠 AI-Powered Translation: Uses the Gemma language model.
• 📊 Model Evaluation: Measures translation quality with BLEU, chrF, exact match, and latency.

Tech Stack: Python, Gemma, Unsloth, Hugging Face, Gradio, Tesseract OCR, and Google Colab.

Goal: Make multilingual communication simpler by combining AI translation with image text extraction.