Logo Lanfrica

6lxck-Schofield/visual-translator-app

Domain:

natural language processing

Record type:

software
Creator:
6lx
Host:
OCR app that scans text in isizulu and provides an english translation # Visual Translator An Android application that uses computer vision and optical character recognition (OCR) to detect and recognize text from images, then provide translations of the extracted text. The project combines a custom-trained OCR model with an Android application developed in Kotlin using Android Studio. The application uses the device camera to capture text and processes the image through a computer vision pipeline before displaying the recognized text and its translation. ## Features * Capture text using the device camera * Detect text regions within images * Preprocess images to improve OCR performance * Recognize text using a custom-trained OCR model * Translate recognized text * Display OCR and translation results within the Android application * Process images directly on an Android device ## System Architecture The application follows the following processing pipeline: ```text Camera Input ↓ Image Capture ↓ Image Preprocessing ↓ Text Detection ↓ OCR Model ↓ Recognized Text ↓ Translation ↓ Translated Output ``` ## Technologies Used ### Android Application * Kotlin * Android Studio * Android SDK * CameraX ### Computer Vision * OpenCV * Image preprocessing techniques * EAST text detection model ### Machine Learning * Python * Custom-trained OCR model * Model training and evaluation pipeline ## OCR Model The OCR component was developed and trained to recognize text from synthetic generated text images. The model forms the core text-recognition component of the application. The OCR pipeline consists of: 1. Input image acquisition 2. Image preprocessing 3. Text region detection 4. Extraction of text regions 5. OCR model inference 6. Text recognition The trained model is integrated into the Android application to allow text recognition directly from captured images. ## Image Processing Pipeline Before OCR inference, captured images can be processed to improve the quality of the input. The preprocessing pipeline may include operations …