Unsupervised Person Re-Identification via Block Processing and Random Ordering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing person Re-identification (Re-ID) methods face significant challenges due to the high workload required for labeling massive person images, making it unrealizable in many scenarios.
Innovation Solution
The proposed method involves acquiring an unlabeled data set, performing block processing and random ordering to generate out-of-order data and negative sample data, and then using unsupervised learning to obtain an ID network for person Re-ID, thereby reducing the workload and ensuring accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual labeling is used for person images in person Re-ID, then identification accuracy can be maintained, but the workload becomes significant and may even be unrealizable
Solution Approach 1:
The system performs self-service by automatically generating negative sample data through block processing and random ordering of person images without requiring manual labeling. The unsupervised learning framework enables the model to learn from unlabeled data, making the system self-sufficient and eliminating the need for human annotators while maintaining identification accuracy.
Solution Approach 2:
The method performs preliminary actions by pre-processing person images through block processing and random ordering to generate out-of-order data and negative samples before the actual Re-ID task. This preliminary data preparation creates a ready-to-use training dataset that eliminates the need for subsequent manual labeling work.
2Productivity
If unsupervised learning is used to reduce labeling workload, then productivity improves, but the complexity of the processing method increases
Solution Approach 1:
The processing method is segmented into distinct modular steps: block processing that divides images into segments, random ordering that shuffles segments, negative sample generation that creates training data, and unsupervised learning that trains the model. This segmentation makes the complex processing pipeline more manageable and systematic.
Solution Approach 2:
The patent introduces intermediary components including block processing as a mediator between raw images and training data, and random ordering as an intermediary transformation that creates out-of-order data. These intermediaries simplify the overall process by breaking down the complex task of unsupervised learning into manageable intermediate steps.
Data Source
AI summary
Disclosed are a person Re-identification (Re-ID) method, apparatus, and device and a storage medium, and the method includes: acquiring a data set, where pieces of data in the data set are unlabeled person images; performing block processing on each piece of data in the data set, performing random ordering on each piece of blocked data to obtain out-of-order data corresponding to each piece of data, and generating negative sample data corresponding to each piece of data based on each piece of data and the corresponding out-of-order data; and performing unsupervised learning based on each piece of data in the data set, the out-of-order data of each piece of data, and the negative sample data of each piece of data to obtain a corresponding ID network, and performing person Re-ID based on the ID network.