Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A kind of cell detection method

A detection method and cell technology, applied in the field of cell detection, can solve the problems of large fluctuations in loss curves, slow calculation speed, and low accuracy of results, and achieve detection speed and calculation resource consumption, good detection accuracy and detection speed, The effect of small fluctuations in the loss curve

Active Publication Date: 2021-10-15
ARMY MEDICAL UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The above two methods have the defects of low result accuracy, slow calculation speed, large calculation resource consumption, large fluctuation of loss curve and high loss

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
  • A kind of cell detection method
  • A kind of cell detection method
  • A kind of cell detection method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0093] In this embodiment, the original cell picture is cut into 2048*2048, and the specific cell is calibrated by an expert, and the specific type and cell location are marked, stored as an xml file, and the calibrated specific cell picture set is used as a training set for the picture Take training samples;

[0094] 2. Perform statistics and analysis on the area, aspect ratio and the number of specific cells in each specific cell in the training sample, and select the most suitable network parameters, which mainly include the anchor_size parameter and the anchor_scalse parameter;

[0095] 3. Set the network parameters and use the training samples to train the algorithm. Before entering the algorithm, the pictures are uniformly scaled to 600*800, and the training model and training parameters are saved, including loss parameters, accurate parameters, etc.;

[0096] 4. Put the picture of the cell to be detected (cut to 2048*2048) into the trained model for testing, and finally...

Embodiment 2

[0098] In this embodiment, we choose the resolution of the pictures to be 2048*2048, manually mark 98 specific cells in a total of 73 pictures, and select 58 pictures (there are 78 manually marked specific cells in 58 pictures) to pair The neural network is trained and tested with the remaining 15 pictures (20 of the 15 pictures have specific cells manually labeled).

[0099] In this embodiment, the YOLO network, the FASTER network and the multi-task-driven deep detection anchor network provided by the present invention are selected for training and testing. Finally, the loss curves of the three networks are obtained as Figure 4 shown.

[0100] It can be seen from the figure that the YOLO loss curve fluctuates greatly, and even after 1000 iterations, it still has a high loss. The multi-task-driven deep detection anchor network provided by the present invention is obviously better than the YOLO network.

[0101] It can be seen from the figure that the loss curve of the prese...

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 present invention provides a cell detection method, comprising the following steps: S1: counting the deep features of sample data and generating a deep detection anchor network; S2: training the deep detection anchor network; S3: using the trained deep detection anchor network to detect to-be-detected data. Compared with the current cell detection method, the cell detection method provided by the present invention has better detection accuracy and detection speed, and achieves a balance between detection speed and computing resource consumption.

Description

technical field [0001] The invention relates to a cell detection method, in particular to a cell detection method using a multi-task-driven deep detection anchor network. Background technique [0002] Existing cell detection methods include methods based on prior knowledge and methods based on supervised learning. [0003] Methods based on prior knowledge include Hough transform, Gaussian filter Laplacian, voting algorithm based on radial symmetry, etc., which are mainly designed by the designer based on the understanding and design of the task, and may be affected by the subjective preference of the researcher. [0004] Methods based on supervised learning include vector machines, random forests, convolutional neural networks, etc. [0005] Both of the above two methods have the defects of low result accuracy, slow calculation speed, large calculation resource consumption, large loss curve fluctuation and high loss. Contents of the invention [0006] In order to solve t...

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 Patents(China)
IPC IPC(8): G06K9/00G06N3/04G06N3/08
CPCG06N3/08G06V20/698G06N3/045
Inventor 刘丽肖晶晶吴毅翟永平谭立文
Owner ARMY MEDICAL UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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