Object Detection Using Movement Area Analysis for Carried Cargo
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection systems face challenges in accurately identifying multiple objects within the same image, particularly when the appearance and shape of cargo objects vary, leading to reduced detection accuracy and erroneous associations between objects and people.
Innovation Solution
An information processing system that utilizes a movement area around detected objects to determine the presence of additional moving objects, employing image capturing, object detection, area calculation, and determination units to enhance detection accuracy and prevent erroneous associations by analyzing time-series image data and calculating movement status.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If object detection is performed on multiple objects with varying appearances and shapes in the same image, then the detection coverage is improved, but the detection accuracy deteriorates
Solution Approach 1:
The patent segments the detection process into two independent stages: first detecting the person object, then detecting the cargo object separately using movement area analysis. This segmentation allows each detection task to be optimized independently, maintaining high accuracy for both person and cargo detection despite their varying appearances and shapes.
Solution Approach 2:
The patent introduces a movement area as an intermediary concept between person detection and cargo detection. By defining a movement area around the detected person and analyzing pixel movement within this area, the system can indirectly detect cargo objects without directly analyzing their varying appearances, thus maintaining high detection accuracy across diverse cargo types.
2Productivity
If object association is performed based on detection results, then work recognition capability is improved, but erroneous associations increase
Solution Approach 1:
The patent implements feedback by analyzing the movement status of pixels within the defined movement area. The system continuously monitors whether pixels within the movement area exhibit movement patterns consistent with cargo being carried, providing feedback that confirms or refutes the association between detected person and cargo objects, thereby reducing erroneous associations.
Solution Approach 2:
The patent employs dynamic analysis by examining the movement status of pixels over time within the movement area. Instead of static association based solely on spatial proximity, the system dynamically evaluates whether pixels move together with the person, allowing the association to adapt to actual movement patterns and reduce false associations when objects are merely nearby but not actually being carried.
Data Source
AI summary
According to one embodiment, an information processing system includes a processor configured to detect a first object in an image in time-series image data and output object information indicating a detection of the first object. The processor calculates a movement status of one or more pixels included in the image by using a plurality of images in the time-series image data, and then outputs movement information indicating the movement status of the one or more pixels. The processor then detects whether a second object moves in conjunction with the first object based on the object information and the movement information.