Research on digital recognition of bank cards based on CRNN

With the popularity of smartphones and the rapid development of mobile Internet, mobile payment has become the main payment channel, so the payment scenario will involve the binding and identification of bank cards. To address this problem, a bank card number recognition algorithm based on SSD and CRNN is proposed. Firstly, the SSD target detection algorithm is used to locate the bank card number region and obtain the target region. Due to the privacy of bank card numbers, this paper uses a data enhancement method to obtain the training dataset, CNN to extract the features of the images and then RNN to predict the sequences, and finally a translation layer of CTC to obtain the final result. Experiments show that the method can accurately identify bank card numbers and meet the recognition timeliness.