The invention discloses a tire DOT information identification method based on end-to-end
deep learning, and the method comprises the following steps: carrying out the
feature extraction of a tire image, obtaining first feature maps outputted in N stages, and carrying out the
feature fusion of the first feature maps outputted in N stages, and obtaining a second feature map; performing DOT information rough positioning on the fused second feature map to detect whether three characters of DOT and position information thereof exist or not, and obtaining an area map; generating a
mask image from the region image, multiplying the
mask image by the second feature image, performing DOT information fine positioning on a third feature image obtained by multiplying to obtain DOT information text probability and position information, and positioning to candidate text blocks with angles; performing tire bending
direction detection on the first feature map output in the last stage to obtain character
direction information of the tire
tread; performing affine transformation on the candidate text blocks and the character
direction information of the tire
tread, and converting the candidate text blocks and the character
direction information into horizontal text blocks in the upward direction; and inputting the horizontal text blocks into a
text recognition network based on
deep learning to carry out DOT
character recognition to obtain final recognition information.