Image-Capture Feature Detection for Low-Bandwidth Target Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing surveillance systems that determine the presence of a target subject by analyzing captured images increase communication data volume, especially when multiple image-capturing apparatuses are used, leading to increased network load.
Innovation Solution
An information processing system that encrypts captured-image data using photoelectric random numbers to generate encryption keys, performing encryption at the image-capturing apparatus level to reduce communication data volume and enhance confidentiality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If image recognition processing is performed on images captured by a monitor camera to identify operators and their states, then operator monitoring capability is improved, but processing time and system complexity increase
Solution Approach 1:
The system performs preliminary actions by capturing operator images in advance and preparing recognition models before actual monitoring is needed. The image recognition processing is pre-configured with multiple models (face recognition, body recognition, state recognition) so that when monitoring is required, the system can quickly process images without delay for model preparation or configuration.
2Measurement precision
If multiple recognition models (face, body, state) are used to comprehensively monitor operators, then monitoring accuracy is improved, but device complexity increases
Solution Approach 1:
The monitoring system is segmented into three independent recognition modules: face recognition model, body recognition model, and state recognition model. Each model focuses on a specific aspect of operator monitoring (identity, posture, state), allowing the system to achieve comprehensive monitoring accuracy while maintaining manageable complexity through modular design. Each segment can be independently trained, deployed, and updated.
Solution Approach 2:
The image recognition system is designed with multi-functionality to handle multiple monitoring tasks simultaneously. The same base image processing infrastructure supports face recognition, body recognition, and state recognition through different recognition models, reducing overall system complexity compared to having separate dedicated systems for each function.
3Measurement precision
If high-resolution images are captured and processed for accurate recognition, then recognition accuracy is improved, but processing time and computational resources increase
Solution Approach 1:
The system applies local quality processing by focusing computational resources on critical regions of the image. The face recognition model concentrates on facial features, the body recognition model on body posture regions, and the state recognition model on relevant state indicators. This localized processing approach maintains high recognition accuracy while reducing overall processing time compared to analyzing entire high-resolution images uniformly.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
To reduce a volume of communication data necessary for determination performed by a system that determines whether there exists a target subject, on the basis of a captured image. An information processing system according to the present technology is an information processing system that includes an image-capturing apparatus and an information processing apparatus, the image-capturing apparatus including an image-capturing section that is used to obtain a captured image of a subject, and a feature detector that detects, on the basis of the captured image, feature information that is information indicating a feature of the subject, the image-capturing apparatus or the information processing apparatus including a determination section that determines whether there exists a target subject in the captured image on the basis of detected feature information that is the feature information detected by the feature detector, and on the basis of target feature information that is the feature information to be targeted, the target subject being a subject for which the detected feature information matches or is similar to the target feature information.