CNN-optimized text recognition with binary embeddings for Arabic expiry date recognition

Journal of Electrical Systems and Information Technology

Table 2 Summary of Approaches on the Expiry Date Recognition Methods

Approach	Dataset (Synthetic/ Real)	Image Format	Image Content	Latin/ Arabic	Filled in/ Dot-matrix	Inference Speed (img/s)	Recognition Approach	Test Accuracy (%)
Florea and Rebedea [5]	Both	Colored	Expiry Date	Latin	Both	0.25	RCNN	72.7
Ashino and Takeuchi [2]	Real	Binarized	Single Digit	Latin	Dot-matrix	–	DNN	90
Khan [11]	Real	Colored	Single Digit	Latin	Both	0.89	CNN	90
Gong et al. [6]	Real	Colored	Expiry Date	Latin	Dot-matrix	–	CRNN	95
Seker and Ahn [20]	Both	Colored	Expiry Date	Latin	Both	0.12	DAN	97.74
Ours	Synthetic	Colored	Expiry Date	Arabic	Dot-matrix	1 k	CNN with Binary Embedding	98.94

The highlighted information in bold demonstrates the significance of our work on the expiry date recognition in Arabic dot-matrix format.