Target detection method based on incremental learning and automatic driving method

A target detection and incremental learning technology, applied in neural learning methods, instruments, biological neural network models, etc., can solve problems such as model performance degradation and limit model learning, and achieve high accuracy, good practicability, and high reliability. Effect

Pending Publication Date: 2022-07-01
CENT SOUTH UNIV
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this approach still cannot avoid the problem of forgetting, and it faces a dilemma-increased loss will limit the ability of the model to l

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 incremental learning and automatic driving method
  • Target detection method based on incremental learning and automatic driving method
  • Target detection method based on incremental learning and automatic driving method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0048] like figure 1 Shown is a schematic diagram of the method flow of the target detection method of the present invention: the incremental learning-based target detection method provided by the present invention includes the following steps:

[0049] S1. Obtain the original target detection initial model; specifically, a neural network-based target detection model that can abstract the structure into the form of P(x)=d(f(.)); where f(.) is a feature extractor, used for Map the image into a feature tensor of a specific dimension; d(.) is the detection head, which is used to decode the feature into a set number of target frames and the corresponding category;

[0050] The original target detection initial model specifically includes the Faster-RCNN model, the YOLO model and the Swin-Transformer model;

[0051] S2. Pre-train some parameters of the feature extractor of the original target detection initial model obtained in step S1 to obtain a general target detection feature ex...

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 incremental learning. The method comprises the steps of obtaining an original target detection initial model; pre-training partial parameters of a feature extractor of the original target detection initial model to obtain a universal target detection feature extractor; performing structure expansion on the target detection feature extractor by adopting the detection head and the parameter mask to obtain an expanded target detection model; and performing actual target detection by adopting the expanded target detection model. The invention further discloses an automatic driving method comprising the target detection method based on incremental learning. According to the incremental learning-based target detection method and the automatic driving method provided by the invention, a new incremental learning target detection method is innovatively provided; through innovative addition of a detection head technology and a mask technology, incremental learning target detection is realized, and the method is high in accuracy, high in reliability and good in practicability.

Description

technical field [0001] The invention belongs to the field of computer image processing, and in particular relates to an incremental learning-based target detection method and an automatic driving method. Background technique [0002] With the development of economy and technology and the improvement of people's living standards, target detection technology has been widely used in people's production and life, bringing endless convenience to people's production and life. [0003] Current target detection methods often use deep learning models, and the learning method is mainly through stochastic gradient descent to fit the distribution of the given data. This form makes the performance of the deep learning model heavily dependent on the number of training samples at this stage, and each training process is coverage rather than incremental. Continuing training with new data on an already trained model creates the problem of catastrophic forgetting. This means that once the m...

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): G06V10/25G06V20/56G06V10/774G06V10/776G06V10/82G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/217G06F18/214
Inventor 梁毅雄赵嘉伟刘剑锋刘晴
Owner CENT SOUTH UNIV
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