Book positioning method based on computer vision
A technology of computer vision and positioning method, applied in computer components, computing, energy-saving computing, etc., can solve problems such as interference, increasing book identification obstacles, and inability to assist book grasping, so as to achieve high-accuracy book positioning and improve utilization Efficiency and time-saving effect of looking up data
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0064] This embodiment mainly introduces a computer vision-based book positioning method. For the specific method, please refer to figure 1 .
[0065] A computer vision-based book positioning method, the method comprising:
[0066] Step S1, taking pictures by a camera, collecting the target area of the book, and storing it as a picture;
[0067] Step S2, build and train a text detection model, mark all texts in the picture with a text area frame;
[0068] Step S3, based on the text area frame, carry out book instance segmentation to the picture, thereby obtaining the book instance and the position information of the book instance in the picture;
[0069] Step S4, constructing and training a text recognition model, identifying the text area frame, and merging the text recognition results in the same book instance to obtain the title recognition result of each book instance;
[0070] Step S5 , matching the title recognition result with the title of the book entered by the r...
Embodiment 2
[0108] Based on the foregoing Embodiment 1, this embodiment mainly introduces another method for locating books based on computer vision.
[0109] A computer vision-based book positioning method, the method comprising:
[0110] Step S0, build and train a text detection model and build and train a text recognition model;
[0111] Step S1, taking pictures with a camera, collecting the target area, and storing it as a picture;
[0112] Step S2, mark all texts in the picture with a text area frame;
[0113] Step S3, based on the text area frame, carry out book instance segmentation to the picture, thereby obtaining the book instance and the position information of the book instance in the picture;
[0114] Step S4, identifying the text area frame, and merging the text recognition results in the same book instance to obtain the title recognition result of each book instance;
[0115] Step S5 , matching the title recognition result with the title of the book entered by the reader...
Embodiment 3
[0162] Based on Embodiment 1, this embodiment further discusses the problems encountered when the application is improved and the improvement method:
[0163] For the ready-made datasets for instance segmentation of dense books in the scenario without bookshelves, the instance segmentation model trained with instance segmentation datasets in other scenarios is not effective. The solution of the present application is to creatively propose a traditional algorithm for book instance segmentation, so as to avoid the problem of low segmentation accuracy due to lack of corresponding data sets. The book instance segmentation algorithm only needs to be based on a neural network for text detection, and the data set for text detection is rich and the detection algorithm is relatively mature, making the overall algorithm easy to implement.
[0164]The placement of adjacent books on the bookshelf is often very close, the textures of these books are similar, and there are various placement...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More