Signal Retrieval Device for Modified Attribute Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image retrieval methods struggle to recognize similar images when attributes such as hairstyle or facial expressions are modified, leading to reduced accuracy in identifying similar images of the same subject.
Innovation Solution
A signal retrieval device and method that modify attribute values of input signals and stored signals using a modification unit, allowing for the retrieval of similar signals by extracting latent variables representing features and generating modified signals using neural networks, enabling the comparison and retrieval of images with modified attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional feature amount extraction is used for image retrieval, then the accuracy of detecting the same image is high, but the accuracy of recognizing similar images of the same subject with modified attributes deteriorates
Solution Approach 1:
The patent segments the image feature representation into two distinct components: identity features (capturing the subject's inherent characteristics) and attribute features (capturing modifiable characteristics like hairstyle, clothing, and expressions). This segmentation is achieved through separate neural network pathways that extract and independently process these feature types, allowing the system to maintain high identity recognition accuracy while being tolerant of attribute variations.
Solution Approach 2:
The patent extracts and isolates the identity feature component from the overall image features using a dedicated identity verification network. This extracted identity feature is then used as the primary basis for retrieval, while attribute features are either excluded or given minimal weight in the similarity calculation, effectively separating the stable identity information from the variable attribute information.
2Adaptability or versatility
If attribute values are modified to retrieve similar signals, then the retrieval of images with varying attributes improves, but the system complexity increases
Solution Approach 1:
The patent creates a simplified copy or representation of the image in the form of extracted feature vectors (identity features and attribute features). Instead of manipulating the original complex image data or generating multiple modified versions of images, the system works with these compact feature representations, which are much simpler to process and compare while retaining the essential information needed for retrieval.
Solution Approach 2:
The patent changes the parameter space from working with raw pixel data or complex image transformations to working with extracted feature vectors. By transforming the problem into this simplified parameter space, the system can efficiently handle attribute variations through simple feature comparison and weighting operations rather than complex image processing.
Data Source
AI summary
A signal retrieval device includes a modification unit and a signal retrieval unit. The modification unit modifies a value of an attribute of a target represented by an input signal or a stored signals stored in a signal storage unit or a value of an attribute relating to a signal generation source of the input signal to acquire a plurality of modified values of the attribute. The signal retrieval unit retrieves a stored signal of the stored signals similar to the input signal using the input signal or the stored signals in which the attribute is modified according to each of the plurality of modified values of the attribute acquired by the modification unit.