Method and device for detecting text

The embodiment of the invention provides a method and device for detecting a text, and belongs to the field of computer vision. The method comprises the following steps: inputting an image of a to-be-detected text into a preset convolutional neural network structure to obtain candidate box information of each candidate box in a plurality of candidate boxes split from the image; for any candidate frame in the plurality of candidate frames, determining whether the candidate frame is located at a left boundary, a right boundary or other positions based on the left probability, the right probability and other probabilities of the candidate frame so as to determine whether the candidate frame is a left frame, a right frame or other frames; aiming at any left frame, determining a paired right frame according to the basic parameters of the left frame; and based on the basic parameters of the left frame and the basic parameters of the paired right frame, determining a line of text area detection frame. Therefore, the time consumption phenomenon is reduced, and the calculation speed is increased.