Seal recognition method and system based on deep learning and storage medium
A technology of deep learning and recognition method, applied in the field of seal recognition method, system and storage medium based on deep learning, can solve the problems of poor seal fan-shaped text correction effect, low text recognition accuracy, and few seal recognition scenes, etc. User satisfaction, reduced review time, and accurate seal recognition results
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0060] as attached figure 1 Shown, the seal recognition method based on deep learning of the present invention, this method is specifically as follows:
[0061] S1. Obtain the image to be recognized, use the target detection model to detect the seal in the image to be recognized, and obtain the coordinate information and the type of the seal corresponding to the seal in the image to be recognized; wherein, the image to be recognized includes a circular seal, an oval seal, a rectangular seal and a square Seal; scenes to be recognized images include bills, contracts, licenses and red-headed documents;
[0062] S2. According to the coordinate information corresponding to the seal in the image to be recognized, cut out the image of the seal to be recognized;
[0063] S3. Use the image rotation angle classification model to identify the rotation angle of the image of the seal to be recognized, and correct the image of the seal to be recognized according to the rotation angle;
[...
Embodiment 2
[0093] as attached figure 2 Shown, the seal recognition system based on deep learning of the present invention, this system comprises,
[0094] The seal target detection module is used to obtain the image to be recognized, detect the seal in the image to be recognized through the target detection model, obtain all the seal images in the image to be recognized, and identify the corresponding type of each seal and the corresponding coordinates of the seal in the image to be recognized information; and according to the coordinate information corresponding to the seal in the image to be recognized, cut out the seal image to be recognized; wherein, the seal type includes a circular seal, an oval seal, a rectangular seal and a square seal;
[0095] The seal tilt correction module is used to use the image rotation angle classification model to identify the rotation angle of the seal image to be recognized, and correct the seal image to be recognized according to the rotation angle; ...
Embodiment 3
[0105] Example 3: Taking the letter of introduction stamped with the special seal of the Railway Engineering Headquarters Hospital as an example
[0106] (1) Obtaining the image of the letter of introduction as the image to be identified;
[0107] (2) Use the target detection model YOLOv5 to detect the seal in the image to be recognized obtained in step (1); specifically: after inputting the image to be recognized into the YOLOv5 model, the YOLOv5 model predicts the coordinate pixel information of the seal in the image, namely Generate the bounding box of the seal in the introduction letter, the bounding box is the x, y pixel coordinates of the upper left corner and the lower right corner of the seal in the image; and recognize that the seal type in the image to be recognized is a circular seal;
[0108] (3), according to the stamp pixel coordinate information of the image to be identified that is acquired, cut out the image of the seal to be identified on the image to be iden...
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com