Classification of Bread Baking States Using YOLOv4 Technique

Baking is a complex process that encompasses various stages, involving heat and mass transfer. During baking, dough undergoes complex chemical and physical changes, including water evaporation, volume expansion, starch pasting, and protein denaturation. These changes are interrelated, and sometimes a change in one stage affects the following processes. Traditional baking techniques rely heavily on the expertise of experienced bakers. However, with the looming challenge of labor shortage in the future, industrial development is likely to be difficult. In this study, we automated the roasting process using image recognition technology. The YOLOv4 technique was used to automatically categorize the baking time stages of Taiwanese buns. Object size and pixel brightness features were extracted from the images in the dataset using the established CNN model. These images were obtained from video recordings of the baking process. The developed method showed promising results with an accuracy of 94.8% in determining the bread baking time intervals. This deep learning model has the potential to extend its application to various types of baked goods. An AI aid can be available for baking, especially in the absence of expensive manual labor. The implementation of AI in the bakery industry also contributes to sustainable management,