Computer Vision Image Retention for Retail Subject Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high cost of storing video data from high-resolution cameras in retail stores, particularly for extended periods, and the lack of effective methods to identify subjects in low-quality images for crime prevention.
Innovation Solution
A computer vision system that selectively retains high-resolution images of subjects' facial and bodily features in non-volatile memory, reducing the amount of stored data by identifying and saving images with the most physical features, and tailoring data retention based on specific store needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution video data is stored for extended periods, then subject identification capability is improved, but storage cost increases
Solution Approach 1:
The system extracts and retains only the most valuable image data - specifically, the highest-resolution image capturing the subject's physical features (face, body, distinguishing characteristics) - while discarding redundant sequential frames. This selective extraction maintains subject identification capability while dramatically reducing storage requirements
Solution Approach 2:
The system discards redundant video data that provides no additional identification value, while recovering and preserving the essential identification information in a compact form. The highest-quality image is retained for future identification needs, eliminating the need to store entire video sequences
2Measurement precision
If all sequential image data is retained, then subject identification accuracy is improved, but data processing complexity increases
Solution Approach 1:
The system performs preliminary analysis during video capture to identify and flag the highest-quality image containing the subject's physical features. This advance preparation eliminates the need for complex post-processing of entire video sequences, as the identification-critical frame has already been selected and marked for retention
3Quantity of substance
If high-resolution images are selectively retained, then storage efficiency is improved, but risk of losing identification-quality images increases
Solution Approach 1:
The system incorporates feedback mechanisms that continuously monitor image quality metrics and subject detection status. When a subject is detected, the system evaluates image quality and provides feedback to retain the highest-resolution capture. This closed-loop approach ensures identification-quality images are preserved while maintaining storage efficiency
Solution Approach 2:
The system dynamically adjusts retention parameters based on detected subject characteristics and image quality. Rather than uniformly retaining all frames or using fixed resolution thresholds, the system adapts retention criteria to preserve only those images that meet identification quality standards, optimizing both storage efficiency and reliability
Data Source
AI summary
Systems and methods of performing image data retention associated with a computer vision system are described. In one exemplary embodiment, a method is performed by a network node operationally coupled to a set of optical sensor devices positioned about a retail space. Further, each optical sensor device has a field of view towards a certain region about the retail space and is operable to capture sequential images that correspond to the certain region. The method includes storing, in non-volatile memory, data that corresponds to the sequential images captured the optical sensor devices of a subject as that subject traverses about the retail space. The method also includes selecting a portion of the stored image data that corresponds to a physical view of the subject or a certain activity performed by the subject while about the retail space to retain in the non-volatile memory.