Abstract

The growth of new trading card games and systems has created a need for more available tools to identify and catalogue card collections. Similar identification systems have been created as applications tracking a single card system, making new cards difficult. This project addresses the need for a generic and simple detection system capable of identifying and displaying trading cards from various games and trading card packs. By developing a pipeline that uses computer vision and various levels of machine learning techniques. This project is a web-based utility that creates Android SDK to detect trading cards with OpenCV and Python. Unlike highly specific applications tailored to a single card game, this solution provides a generalized pipeline that can be applied to any pack or game. Trading Card Recognition offers a scalable cross-platform tool that enhances the card collection experience for both hobbyists and collectors alike.