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

Target detection model training method and device based on deep learning and storage medium

A technology for target detection and model training, applied in the field of deep learning target detection, can solve problems such as low confidence and wrong category labeling, achieve good performance, and improve accuracy and recall.

Pending Publication Date: 2020-10-16
DONGGUAN ZHENGYANG ELECTRONICS MECHANICAL
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At the same time, there will be some mislabeled categories
When using the classification loss function to calculate the classification loss, the classification loss function only maximizes one category and minimizes other different categories. If similar categories are simply treated as different categories, the confidence of similar categories will decrease. is suppressed, resulting in lower confidence

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 model training method and device based on deep learning and storage medium
  • Target detection model training method and device based on deep learning and storage medium
  • Target detection model training method and device based on deep learning and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0028] First, refer to figure 1 Describe the deep learning-based target detection model training method of the present invention. Such as figure 1 As shown, the deep learning-based target detection model training method includes the following steps:

[0029] S101. Test the training images containing target labels, and obtain the targets contained in each training image. Specifically, when testing the training images containing target labels, the training images can be tested with the help of external public data sets (such as COCO data sets, etc.); it is also possible to directly input the training images into the trained target detection In the model, the training image is tested by the target detection model. At this time, the confidence threshold will be set lower to reduce the possibility of missing some targets. Specifically, the threshold can be set according to the recall rate of the algorithm, such as the confidence level corresponding to the recall rate of 99%.

...

Embodiment 2

[0037] When training an object detection model, it is necessary to calculate the ground truth (either the background or a certain category) for each location in the training image. In the anchor-based algorithm, the intersection over union ratio (Intersection over Union, iou) between the bounding box (anchor) of the target label and the prediction box is usually calculated. If the iou is greater than the preset threshold, it is set as a positive sample, otherwise it is Negative samples. Due to the influence of factors such as the translation and size of the target, the iou of some targets and anchors may be smaller than the preset threshold. If the positive and negative samples are divided directly based on the iou, it will lead to missed detection of positive samples. In the anchor-free algorithm, when it is judged that the size of the target is within the scale range of the layer network, and there is a feature point in the bounding box marked by the target, the feature poin...

Embodiment 3

[0043] In this embodiment, a further design is made on the basis of the second embodiment, which takes the similarity of categories into consideration when calculating the classification loss of positive samples. For example, when the target is a pedestrian, the probability of the cyclist being judged as the background will be lower.

[0044] Specifically, first, set the similarity matrix of each category, the value of each element of the main diagonal in the similarity matrix is ​​1, and the other elements take values ​​​​in the interval [0,1] according to the similarity of the category, and the categories are similar The larger the degree, the smaller the value of the element to reduce the interaction between elements. Then, the classification loss of each category is calculated for each feature point, and finally the product is multiplied by the similarity matrix and summed to obtain the classification loss of the positive sample. For example, the similarity matrix looks l...

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 model training method based on deep learning, and the method comprises the steps of testing training images containing target marks, and obtaining a target contained in each training image; comparing the target with the target label to obtain an unlabeled target and a labeled target contained in each training image; obtaining a total loss value accordingto the loss of the labeled targets and the classification loss of the unlabeled targets; and adjusting network parameters of the target detection model according to the total loss value. According tothe invention, the unlabeled targets in the training images are found out, then the classification loss of the unlabeled targets is calculated, other losses of the unlabeled targets are ignored, the trained target detection model has better performance, and the accuracy and recall rate of target detection are improved. In addition, the invention further discloses an electronic device and a computer readable storage medium.

Description

technical field [0001] The present invention relates to the technical field of deep learning target detection, in particular to a deep learning-based target detection model training method, electronic equipment, and a computer-readable storage medium. Background technique [0002] With the rapid development of deep learning technology, the integrated target detection method based on deep learning has also replaced the traditional feature extraction + feature classification target detection method, which has been widely used in many fields. Since the target detection algorithm based on deep learning has the characteristics of high recall rate and low false detection rate, and it can be processed in real time after network structure optimization on a special hardware platform. In the car collision warning system, the target detection method based on deep learning is also applied to detect the target vehicle to judge the possibility of collision between the target vehicle and 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
IPC IPC(8): G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V2201/08G06V2201/07G06N3/047G06N3/045G06F18/241G06F18/2415
Inventor 顾一新
Owner DONGGUAN ZHENGYANG ELECTRONICS MECHANICAL
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