A generalized framework for recognition of expiration dates on product packages using fully convolutional networks. (1st October 2022)

• Record Type: Journal Article
• Authors: Seker, Ahmet Cagatay; Ahn, Sang Chul
• Is Part Of: Expert systems with applications. Volume 203(2022)
• Publication Date: 2022-10-01
• Subjects: Expiration date -- Recognition -- Date parser -- Food safety -- Convolutional network -- Deep learning; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2022.117310
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: It is important to understand the expiration date. However, it is challenging for machines to understand it. Most previous methods recognize expiration dates in limited conditions. To address this problem, a generalized framework for detecting and understanding expiration dates has been proposed. This framework handles challenging cases and distinguishes 13 different date formats. Unlike previous methods, a neural network-based date parser is adopted in the framework to understand the meaning of an expiration date by identifying the day, month, and year. The experimental results demonstrate the proposed framework achieves 97.74% recognition accuracy for expiration dates in various formats and challenging cases. Since there is no publicly available dataset of expiration dates, a novel dataset collection named ExpDate was created and opened. Highlights: A generalized framework for detecting and understanding expiration dates. A new DMY (Day, Month, Year) detection network for understanding expiration dates. The framework dealing with various expiration date formats and challenging cases. A novel dataset collection of expiry date images, consisting of six datasets.