How I Built a Wheel of Fortune Game with Roboflow

Warren Wiens, a marketing strategist exploring AI, developed a project using computer vision to enhance the "Wheel of Fortune" viewing experience by displaying both correct and incorrect letter guesses in real time. By connecting a USB HDMI capture device to a Roku express, Wiens captured images from the show to train a model capable of detecting letters on the game board and those called by players. Two separate models were created to manage the different letter types, with images processed through Roboflow for training and inference. The project involved setting up a local Roboflow inference server to handle real-time video analysis, along with a web interface using Flask and SocketIO to display the updated puzzle board and called letters. The Python-based code, available on GitHub, offers a detailed look at the process, demonstrating the integration of technology in a traditional game setting.