A Transformer-Based Approach for Fruit Spoilage Identification

One of the most significant problems in the fruit industry is fruit spoilage, which creates categories of losses across all fruit types. The main objective of the current study is to develop a transformer-based system for identifying spoilage of fruit and reducing waste. It employs vision transformer (ViT) to identify fresh and spoiled fruit categories among eight types of fruit. The experimental findings indicate that transformer-based fruit spoilage identification system can attain a high performance without the need for destructive analysis. With a high accuracy (99.89%) attained in distinguishing fresh fruit from spoiled fruit, the proposed system may be useful for improving real-time fruit quality and spoilage assessments, thereby improving food management efficiency.