An automatic taxi receipt text recognition application system
Autor: | Hang Du, Mengqi Wu, Yang'an Zhang, Yakun Cui, Xueguang Yuan, Liu Weiliang |
---|---|
Rok vydání: | 2020 |
Předmět: |
Receipt
050210 logistics & transportation Artificial neural network Computer science 05 social sciences Real-time computing Process (computing) Image processing 02 engineering and technology Communications system Convolutional neural network Task (computing) 0502 economics and business 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Data set (IBM mainframe) |
Zdroj: | BMSB |
Popis: | Gathering and managing the data of taxi receipts in reimbursement process consumes a lot of manpower. Due to the bad painting and the irregular arrangement of the taxi receipts, the automatic taxi receipt character recognition task has not been solved yet. This paper proposes an automatic taxi receipt text recognition application system based on a recognition system and a communication system. In the recognition system, a Single Shot Multi-Box Detector(SSD) neural network is used to determine the data region, and a CNN (convolution neural network)-GRU (gated recurrent unit) combined neural network is designed for character recognition. For the networks, our data set contains 1800 images of taxi receipts, and the cross validation results show that the accuracy reaches 94.36%. The communication system is deployed on a multimedia server and the system can be used on both computers and phones. |
Databáze: | OpenAIRE |
Externí odkaz: |