Video Analysis for Abnormal Behavior Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video monitoring technologies fail to accurately determine abnormal behavior due to limitations in analyzing depth relationships between objects and persons in two-dimensional space, leading to incorrect assessments of individuals performing dangerous actions.
Innovation Solution
An information processing program and device that analyze video frames to accurately identify interactions between objects and persons by using a machine learning model to specify regions and relationships, generating attention maps to contextualize object and person interactions, and determining abnormal behavior based on these relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If two-dimensional Bbox positional relationship analysis is used, then the system is simple and fast, but depth analysis cannot be performed leading to incorrect abnormal behavior detection
Solution Approach 1:
The patent transitions from two-dimensional Bbox analysis to three-dimensional spatial analysis by introducing depth information through camera calibration parameters and 3D coordinate transformations. This allows the system to determine whether a person is actually on top of an object or merely positioned behind it in the 2D image plane, thereby resolving the false positive problem while maintaining computational efficiency through pre-calibrated camera parameters.
2Productivity
If machine learning models are used for object detection, then detection speed is improved, but false positives occur due to inability to distinguish depth relationships
Solution Approach 1:
The patent introduces an intermediary processing layer that takes the 2D Bbox coordinates from the machine learning detection model and transforms them into 3D spatial relationships using camera calibration data. This intermediary step acts as a bridge between fast 2D detection and accurate 3D reasoning, maintaining the speed advantage of ML models while adding the depth discrimination capability needed to eliminate false positives.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An information processing program causes a computer to execute processing including: acquiring a video; specifying a first region that includes an object included in the video, a second region that includes a person included in the video, and a relation that identifies an interaction between the object and the person, by analyzing the acquired video; determining whether or not the person included in the second region performs abnormal behavior, based on the specified first region and the specified relation; and notifying an alert related to appearance of the person who performs the abnormal behavior in a case of determining that the person performs the abnormal behavior.