Owligator is a mobile-based image uploader and label analyzer designed for scanning and extracting information from wine bottle labels using GPT-4 Vision.

Project Details

Owligator is a Flutter-based application that enables users to capture images of wine bottles directly from their phones and upload them to a Flask backend for processing. Once uploaded, the image is analyzed using GPT-4 Turbo's Vision API to extract key details such as wine name, producer, vintage year, type, region, and ABV. The backend performs the analysis asynchronously and saves results locally as JSON entries for future use.

Technologies Used

• Flutter (Dart)
• Flask (Python)
• OpenAI GPT-4 Vision API
• Threading and File I/O in Python

Challenges and Solutions

One major challenge was ensuring reliable network communication between the Flutter app and the Flask server, especially on restricted Wi-Fi networks. To address this, the server was configured to run on all IPs (0.0.0.0) and timeout handling was added to make the app more forgiving to delays. Another challenge was parsing GPT-4's image analysis output into structured JSON; this was solved using a consistent key-value extraction strategy with fallback error handling.

Outcome

The project successfully increased the image recognition range to over 10 feet and enabled rapid background analysis of wine bottle images with persistent logging. It significantly streamlines the wine data entry process and provides a practical demonstration of combining mobile frontend, AI, and backend processing.