Tire DOT information identification method based on end-to-end deep learning
A technology of deep learning and information recognition, applied in the field of image recognition, can solve the problems of low recognition accuracy and achieve the effect of improving detection accuracy, accurate recognition, and improving accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0064] Such as figure 1 Shown, a kind of tire DOT information identification method based on end-to-end deep learning, described method comprises steps as follows:
[0065] S1: The tire image of the tire DOT information collected by the image acquisition hardware system, such as figure 2 As shown, its camera shoots a part of the tire each time, with high resolution, and the characters on the tire can be clearly displayed in the image. Depend on figure 2 It can be seen that the direction of tire bending is different, so the text information can be corrected to a positive direction according to this information. Each set of DOT information will have three characters "DOT" in front, so you can judge whether there is complete DOT information in the image based on this information.
[0066] In this embodiment, feature extraction is performed on tire images with tire DOT information collected to obtain first feature maps output by N stages, and at the same time, feature fusion ...
Embodiment 2
[0131] Based on the tire DOT information recognition method based on end-to-end deep learning described in Embodiment 1, this embodiment also provides a tire DOT information recognition device, the device includes a feature extraction network module, a DOT information rough positioning module, DOT information fine positioning module, tire bending direction detection module, ROI Rotate module and DOT character recognition module;
[0132] The feature extraction network module is used to perform feature extraction on the tire image with the tire DOT information collected, to obtain the first feature map output by the N stages, and to perform feature fusion on the first feature map output by the N stages to obtain the first feature map. Two feature maps;
[0133] The DOT information rough location module is used to roughly locate the DOT information on the second feature map, and is used to detect whether there are three characters of "DOT" and their position information, so as t...
Embodiment 3
[0139] A computer system includes a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the computer program, the steps of the method are as follows:
[0140] S1: Perform feature extraction on the tire image with the tire DOT information collected, respectively obtain the first feature map output by N stages, and at the same time perform feature fusion on the first feature map output by N stages to obtain the second feature map;
[0141] S2: Roughly locate the DOT information on the fused second feature map to detect whether there are three characters of "DOT" and their location information, so as to obtain the area map;
[0142] S3: Generate a mask map from the area map, multiply it with the second feature map, perform DOT information fine positioning on the third feature map obtained after multiplication, and obtain the text probability and position information of the DOT information, so as to locate an a...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More