CNN-based detector, image detection method and terminal

A detector and one-way technology, applied in the field of detection, can solve the problems of high computational complexity of image features, difficulty in applying mobile terminals or embedded devices, and long time consumption

Active Publication Date: 2020-07-07
SPREADTRUM COMM (TIANJIN) INC
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The existing image convolution method for extracting image features has high computational complexity and takes a long time, making it difficult to apply to mobile or embedded devices

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
  • CNN-based detector, image detection method and terminal
  • CNN-based detector, image detection method and terminal
  • CNN-based detector, image detection method and terminal

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0039] As mentioned earlier, CNN-based deep learning methods are widely used in the field of object detection and can be divided into two categories:

[0040] 1. The first category is the method based on the object candidate window, the typical representative is Faster R-CNN. The main principle is to use the Region Proposal Network (RPN) to calculate several object candidate windows on the shared convolutional feature layer; then classify and regress the feature information in the object candidate windows to obtain object category information and position information to complete the object detection task.

[0041] 2. The second category is the candidate window-independent method, typical representatives are YOLO detector and SSD. This type of method does not require additional calculation of object candidate windows and the corresponding feature resampling process. Instead, several anchor windows (Anchor Box) with different scales and aspect ratios are preset directly in 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 CNN-based detector, an image detection method and a terminal. The detector comprises a basic network and a feature extraction module, wherein the basic network comprises a first sub-network and a second sub-network; the first sub-network comprises a plurality of groups of first repeating modules, the data of the previous group of first repetition modules is output to thenext group of first repetition modules, each group of first repetition sub-modules comprises a separation series module for series operation and a first transmission module for transmitting operation,and the first sub-network outputs a first path of feature layer data to the second sub-network; and the second sub-network comprises a plurality of groups of second repetition modules, data of the previous group of second repetition modules is output to the next group of second repetition modules, and each group of second repetition sub-modules comprises the separation series module and a secondtransmission module used for transmitting operation. The scheme of the invention can improve the feature extraction precision, and be more opportunistic to be suitable for mobile terminals or embeddeddevices.

Description

technical field [0001] The present invention relates to the technical field of detection, in particular to a CNN-based detector, an image detection method and a terminal. Background technique [0002] Object detection is to analyze image or video data to determine whether there are certain objects (such as pedestrians, cars or various commodities, etc.) and give the specific positions of these objects. Object detection is a key technology in the field of computer vision. It is widely used in security monitoring, autonomous driving, and intelligent hardware. It is the prerequisite for subsequent high-level tasks such as behavior analysis and semantic analysis. [0003] Among the traditional object detection methods, the most influential ones are the Ada-Boost CascadedModel (Ada-Boost CascadedModel) and the Deformable Part-based Model (DPM). The former is mainly suitable for face detection, while the latter is successfully applied to pedestrian detection, but its detection ac...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04
CPCG06N3/045G06F18/214
Inventor 刘阳罗小伟林福辉
Owner SPREADTRUM COMM (TIANJIN) INC
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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