Pedestrian Re-Identification Network Training via Pseudo-Labeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing pedestrian re-identification network training process is slow and inefficient, resulting in low operation efficiency and poor robustness and applicability.
Innovation Solution
A method that involves acquiring a first-type pedestrian image without a label, producing label information using template matching, training a target pedestrian re-identification network, discarding a target region to create a second-type image, and re-training the network using both image types to increase training samples and improve robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional pedestrian re-identification network training is used, then the network can be trained with existing methods, but the training process is slow and operation efficiency is low
Solution Approach 1:
The patent applies preliminary action by pre-processing pedestrian images to generate pseudo-labels and pre-training the network on these prepared data before final training. The image processing module pre-extracts features and generates labels in advance, so that when the network training begins, the data is already optimized and ready, significantly reducing the overall training time and improving efficiency
Solution Approach 2:
The patent uses copying by creating pseudo-labels that copy the essential information from original pedestrian images. Instead of requiring manual annotation of every pixel and feature, the system generates synthetic labels that replicate the identification information, allowing the network to train on these copied labels which are much faster to produce while maintaining training effectiveness
2Reliability
If traditional pedestrian re-identification network training is used, then the network can perform basic identification, but the robustness and applicability are poor
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting training parameters and data processing methods based on different scenarios. The system can change label generation parameters, image processing parameters, and network architecture parameters to adapt to different pedestrian re-identification tasks, making the network more robust across various applications and conditions
Solution Approach 2:
The patent achieves universality by designing a multi-functional training system that can handle various pedestrian re-identification scenarios. The image processing module can process different types of images, the pseudo-label generation can adapt to different label formats, and the network architecture can be configured for different applications, making the same basic system applicable to multiple purposes
Data Source
AI summary
Provided in the present application are a disk processing method and system, and an electronic device. The method comprises: when disk alarm information is detected, marking a corresponding alarm disk as a faulty disk; detecting the state of a disk group corresponding to the faulty disk; if the state of the disk group is degraded, marking the faulty disk as an isolation disk, and generating alarm information; if the state of the disk group is healthy, determining whether there is a redundant disk group; if there is no redundant disk group, operating the faulty disk according to a first preset rule, and generating alarm information; and if there is a redundant disk group, detecting the state of the redundant disk group, if the state of the redundant disk group is healthy, marking the faulty disk as the isolation disk, and generating alarm information, otherwise, operating the faulty disk according to a second preset rule, and generating alarm information.