ExpirySense: Expiry Date Detection and Proactive Alerts

Abstract

An ExpirySense is an edge-deployable system which is designed to detect the Expiry dates on the product packaging and generate a proactive alert to enhance consumer safety and product lifecycle monitoring. It uses a combination of MobileNet and PaddleOCR to detect and understand date information, even under poor lighting conditions and curved surface designs. This system employs a Region of Interest (ROI) selection logic, identifying patterns around keywords such as MFG and EXP to locate date text. It achieves a detection accuracy of 97.7% on a custom dataset of over 5,000 images, with a precision of 96.2%, recall of 94.8% and the F1-score of 95.5% including curved, dotted, and skewed text formats. Inference time is reduced by 25%, achieving 25–30 ms per image and the system supports multi-language recognition (6 languages). Also, the model size is 9.4 MB which is reduced by 40% with respect to existing models like PaddleOCR v3, easyOCR.

Introduction

Detecting expiry dates on consumer products is a critical challenge in retail, as misreading or ignoring them can lead to health risks, legal issues, and food wastage. Traditional OCR models struggle with curved, rotated, low-contrast, or multi-language text, achieving only 75–92% accuracy and requiring high computational resources, making them impractical for small retailers. Existing solutions like barcode scanning, RFID, or Amazon Go are costly and cannot handle mislabeling or damaged labels.

The proposed ExpirySense system addresses these challenges by using a lightweight OCR-based approach (MobileNet + PaddleOCR) capable of detecting expiry and manufacturing dates in real time, even on curved, skewed, dotted, or compact text formats. It supports multi-language recognition, reduces inference time by 25%, and generates proactive alerts to help retailers and consumers manage products safely. The system uses ROI selection to focus on relevant date regions, improving detection accuracy and reliability across diverse product packaging.

Conclusion

ExpirySense – Expiry date detection and proactive alert is a highly effective approach to track products expiry related info in retail environment. By integrating PaddleOCR model for expiry date detection and alert generation, it achieved the high accuracy for expiry date in dotted printing like case. It has achieved overall 97.7% accuracy for detecting expiry date and their respective manufacturing dates to solve the overhead in retail store. Furthermore, the extracted info given to MobileNET and CRNN which is acting as an alert generation engine here it will decide based on input the alert will be generated or not. The combination of all this advanced notification techniques for expired products. ExpirySense ensures that it can handle real-world situations effectively in retail environment. We used different types of input images products with different types of fonts, but with the same text position, orientation and black shade intensity. The results can be improved by upgrading the dataset. In future work we will try to tackle other problems such as different product name with smaller area, shape and intensity of the expiry date text area. For the improvement of the robustness of this system we will try to enhance our solution to work in any uncontrolled environment.

References

[1] Rani, S. Jansi, and N. Saranya. \"Medical Prescription Analysis Using Machine Learning.\" In 2025 International Conference on Machine Learning and Autonomous Systems (ICMLAS), pp. 376-379. IEEE, 2025. [2] Lu, Yangyu, Yuru Chen, and Shibiao Zhang. \"Research on the Method of Recognizing Book Titles Based on Paddle OCR.\" In 2024 4th International Signal Processing, Communications and Engineering Management Conference (ISPCEM), pp. 1044-1048. IEEE, 2024. [3] Song, Qinghua, Yajun Liu, Haoyue Sun, Yong Chen, and Zheng Zhou. \"Multi-Device Universal Automation Data Acquisition and Integration System.\" IEEE Access (2024). [4] Honari, Mohammad Mahdi, Hossein Saghlatoon, Rashid Mirzavand, and Pedram Mousavi. \"An RFID sensor for early expiry detection of packaged foods.\" In 2018 18th International Symposium on Antenna Technology and Applied Electromagnetics (ANTEM), pp. 1-2. IEEE, 2018. [5] Muresan, Mircea Paul, Paul Andrei Szabo, and Sergiu Nedevschi. \"Dot matrix OCR for bottle validity inspection.\" In 2019 IEEE 15th international conference on intelligent computer communication and processing (ICCP), pp. 395-401. IEEE, 2019. [6] Raj, A. Nirmal, G. Santhosh Kumar, Vellayan Srinivasan, DV Lokeswar Reddy, B. V. Febiyola Justin, and R. Priyadharshini. \"Analysis of Internet of Things based Medication Dispenser Reminder and Drug Expiration Health Monitoring and its Applications.\" In 2024 4th International Conference on Sustainable Expert Systems (ICSES), pp. 241-244. IEEE, 2024. [7] Almutairi, Ahad, Jawza Alharbi, Shouq Alharbi, Haifa F. Alhasson, Shuaa S. Alharbi, and Shabana Habib. \"Date fruit detection and classification based on its variety using deep learning technology.\" IEEE Access (2024). [8] Gong, Liyun, Mamatha Thota, Miao Yu, Wenting Duan, Mark Swainson, Xujiong Ye, and Stefanos Kollias. \"A novel unified deep neural networks methodology for use by date recognition in retail food package image.\" Signal, Image and Video Processing 15 (2021): 449-457. [9] Song, Qinghua, Yajun Liu, Haoyue Sun, Yong Chen, and Zheng Zhou. \"Multi-Device Universal Automation Data Acquisition and Integration System.\" IEEE Access (2024). [10] Muresan, Mircea Paul, Paul Andrei Szabo, and Sergiu Nedevschi. \"Dot matrix OCR for bottle validity inspection.\" In 2019 IEEE 15th international conference on intelligent computer communication and processing (ICCP), pp. 395-401. IEEE, 2019. [11] Lu, Yangyu, Yuru Chen, and Shibiao Zhang. \"Research on the Method of Recognizing Book Titles Based on Paddle OCR.\" In 2024 4th International Signal Processing, Communications and Engineering Management Conference (ISPCEM), pp. 1044-1048. IEEE, 2024. [12] Ma, Chenbo, Fang Li, Zichen Wang, Yaozhong Zhang, and Fengxiang Li. \"Application and Practice of PaddleOCR Based Image Recognition Technology in Business License.\" In 2024 2nd International Conference on Signal Processing and Intelligent Computing (SPIC), pp. 741-745. IEEE, 2024. [13] Rani, S. Jansi, and N. Saranya. \"Medical Prescription Analysis Using Machine Learning.\" In 2025 International Conference on Machine Learning and Autonomous Systems (ICMLAS), pp. 376-379. IEEE, 2025. [14] Karanth, KV Ujwal, A. T. Sujan, YR Thanay Kumar, SudhanvaS Joshi, KP Asha Rani, and S. Gowrishankar. \"Breaking Barriers in Text Analysis: Leveraging Lightweight OCR and Innovative Technologies for Efficient Text Analysis.\" In 2023 2nd International Conference on Automation, Computing and Renewable Systems (ICACRS), pp. 359-366. IEEE, 2023. [15] Ponte, Eric, Xavier Amparo, Kuan Huang, and Daehan Kwak. \"Automatic Pill Identification System based on Deep Learning and Image Preprocessing.\" In 2023 Congress in Computer Science, Computer Engineering, & Applied Computing (CSCE), pp. 1-6. IEEE, 2023. [16] Selvam, Prabu, and Joseph Abraham Sundar Koilraj. \"A deep learning framework for grocery product detection and recognition.\" Food Analytical Methods 15, no. 12 (2022): 3498-3522.

Copyright

Copyright © 2025 Jyoti Madake, Krushna Kalamkar, Prathmesh Katepallewar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.