Authors: Liu Rong; Xu Chengshen; Huang Xiao; Li Lin
Description: In this task, we have used the end-to-end detecting algorithm Yolov3 to accurately localizing the company, the date, the address and the total amount of money with four regions. we utilize the end-to-end recognizing algorithm CRNN to recognizing the company and the address. If the width/height ratio of an image is larger than 12, we enlarge the width of the image to 480. The CRNN network consists of eight convolutional layers, two LSTM layers and a fully connected layer. Each convolutional layer has a batch normalization and a pooling operation followed. If the width/height ratio of an image is smaller than 12, we enlarge the height of the image to 40. After that we extend the image to the same size 40*480 with zero paddings. We utilize Yolov3 again to recognize the date and the amount of money. To gain higher precision, we use specific regular expressions to match the date and amount of money.