SignFlow

Breaking Barriers Between Speakers and Signers

SignFlow is a cross-platform communication app designed to bridge the gap between speakers and sign language users in real-time.
By leveraging AI-powered sign language recognition and speech-to-text technology, SignFlow enables seamless, inclusive conversations between people who communicate differently.

Whether you’re in a meeting, a classroom, or a casual chat, SignFlow ensures that no one is left out of the conversation.

---

📜 Overview

SignFlow provides two-way communication between speakers and signers without delays. It uses a combination of computer vision, natural language processing, and speech recognition to make conversations natural and accessible.

---

✨ Key Features

• Real-time Sign Language Recognition
  Detects and interprets sign language gestures into text and speech instantly using the device’s camera.

• Speech-to-Text Conversion
  Converts spoken words into accurate, readable text for sign language users.

• Two-Way Translation
  Enables back-and-forth communication between speakers and signers in real time.

• Multi-Language Support
  Supports multiple spoken and signed languages for global accessibility.

• Offline Mode
  Works without an active internet connection for uninterrupted communication.

• Customizable Interface
  Large text display, adjustable colors, and accessible controls for all users.

---

🛠 How It Works

1. Select Communication Mode

   • Speaker Mode – For spoken language users.
   • Signer Mode – For sign language users.
   • Mixed Conversation – Both participants can speak/sign.

2. Input Communication

   • Signers: Use your device’s camera to capture hand gestures. The app interprets them into text/speech.
   • Speakers: Speak into the device’s microphone. The app converts speech into text for the signer.

3. View and Respond

   • All responses appear in real time, ensuring a natural conversation flow.

---

📂 Dataset & Model

• Dataset: Sourced from Kaggle for sign language gesture recognition.
• Model: Trained using deep learning with TensorFlow/Keras.
• Note: The .h5 trained model file is not included in this repository due to GitHub file size limitations.
• Instead, the repository contains the training scripts so you can retrain the model locally.