High-dimensional image target defect detection model based on axial self-attention
一种目标缺陷、检测模型的技术,应用在计算机视觉与机器学习领域,能够解决无法高效处理高维高分辨率图片等问题
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Examples
Embodiment Construction
[0032] The specific steps of the present invention will be explained below with reference to the accompanying drawings.
[0033] In step (1), the self-attention layer based on the axial multi-head is constructed, and the self-attention calculation in a specific direction is performed on the width and height, and the mask block is predicted by constructing an inner decoder, so as to promote the encoding The machine learns the global representation of the picture, encodes the key, value, and query, and the prediction target can be expressed as:
[0034]
[0035] In step (2), after the global feature map obtained in (1), anchors of k scales are selected for each d-dimensional point, and a binary classification code of positive and negative samples is constructed for each anchors, and the dimension is changed from d Go to 2*k; define (A x ,A y ,A w ,A h ) four offsets, where (A x ,A y ) is the coordinate of the upper left point in the feature map corresponding to the anch...
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.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap