Supercharge Your Innovation With Domain-Expert AI Agents!

Target detection method based on rapid neural architecture search

A target detection and search algorithm technology, applied in the field of computer vision, to achieve good transferability, short search time, and improved search efficiency

Pending Publication Date: 2021-03-09
NORTHWESTERN POLYTECHNICAL UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Compared with traditional artificial structure design, the design of target detection network based on neural architecture search technology is more challenging

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Target detection method based on rapid neural architecture search
  • Target detection method based on rapid neural architecture search

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment

[0041] 1. Define the search space

[0042] The image features extracted by neural network tend to show the characteristics of reducing fine-grained texture information and increasing high-dimensional semantic information as the number of extraction layers increases. Such as figure 1 As shown, the multi-layer feature fusion module ( figure 1 The FPN in FPN) aims to fuse shallow features with deep features so that texture and semantic information can be preserved at the same time. And the detection head branch module ( figure 1 The Head in ) uses the fused features to perform the classification and regression tasks of the detection frame. The two cooperate with each other to form a complete backend of the general object detection framework. The existing artificially designed fusion modules tend to adopt a regular parallel layer-by-layer connection method, and do not have the characteristic cross-level combination and residual connection capabilities. In order to support the...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a target detection method based on rapid neural architecture search, and the method comprises the steps: defining a search space which consists of a multi-layer feature fusionmodule and a detection head module; constructing a search algorithm; applying a search algorithm to the search space, searching the original image, and detecting to obtain the category and position information of the target in the original image. By using the searching method provided by the invention, the customized network models under different computing scenes can be customized for the targetdetection task in limited time and computing resources, and the balance between the efficiency and the detection precision is considered.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a target detection method. Background technique [0002] The task of object detection has always been a frontier research hotspot in the field of computer vision. Its main purpose is to identify and mark objects of interest in image or video content with boxes. In recent years, with the continuous development of target detection technology, it has been widely used in various fields such as automatic driving, road monitoring, and criminal investigation. At present, most target detection methods use a feature pyramid structure, which is a hierarchical structure, which can make it easier for the neural network model to decode the features extracted by the model during training, and perform hierarchical processing for different types of images. This helps to improve the classification and detection capabilities of the network. However, the manual design of simi...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V10/44G06V2201/07G06N3/045G06F18/253G06F18/214
Inventor 张艳宁张世周高扬王宁
Owner NORTHWESTERN POLYTECHNICAL UNIV
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com