Deep Learning Image Filtering via Feature Vector Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods are inefficient in filtering out similar information, such as images or texts, from vast datasets, requiring manual comparison which is time-consuming and impractical.
Innovation Solution
A deep learning-based method that uses a computing apparatus to acquire and process image data, calculate feature vectors, and compare their similarity, recording new data only when the similarity is below a predetermined threshold, thereby filtering out similar information efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual comparison methods are used to filter similar information, then filtering accuracy can be maintained, but the processing time and labor requirements increase significantly
Solution Approach 1:
The patent replaces manual mechanical comparison with an automated computing system that uses deep learning models to perform image similarity comparison. The computing apparatus automatically extracts features from images using neural networks and calculates similarity metrics, eliminating the need for manual visual inspection while maintaining high accuracy in filtering similar images.
Solution Approach 2:
The patent introduces deep learning models and feature extraction algorithms as intermediaries between the raw image data and the filtering decision. These intermediaries transform images into feature vectors and compute similarity metrics, enabling automated high-accuracy filtering without direct manual comparison.
2Productivity
If deep learning models are used to determine similarity, then filtering efficiency is improved, but computational complexity and resource requirements increase
Solution Approach 1:
The patent segments the image processing task into distinct stages: feature extraction using deep learning models, similarity calculation based on extracted features, and threshold-based filtering decisions. This segmentation allows the system to leverage the power of deep learning for feature extraction while using simpler, more efficient operations for comparison and filtering, thereby improving overall efficiency without excessive complexity.
3Speed
If automated filtering systems are implemented, then processing speed increases, but the complexity of system implementation and maintenance increases
Solution Approach 1:
The patent implements a universal computing apparatus that can handle multiple image filtering tasks using the same deep learning infrastructure. The system is designed to process different types of images and apply various similarity metrics through a unified framework, reducing implementation complexity compared to specialized systems for each specific filtering task.
Data Source
AI summary
According to the present invention, proposed is a method for collecting filtered image data, the method comprising the steps of: obtaining, by a computing apparatus, first image data and recording the first image data in an image data pool; obtaining, by the computing apparatus, second image data; performing, by the computing apparatus, an operation in a deep learning model by using the first image data and the second image data as input values, and calculating a first feature vector corresponding to the first image data and a second feature vector corresponding to the second image data; and comparing, by the computing apparatus, a similarity between the first feature vector and the second feature vector, and when the similarity is less than a certain value, recording the second image data in the image data pool.