Video Rule Creation with Ground-Plane Region Reshaping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Novice users face challenges in creating effective video rules for monitoring systems due to difficulties in accurately defining regions and ground planes, leading to suboptimal performance of video analytics.
Innovation Solution
A system that assists in creating video rules via scene analysis by automatically determining the ground plane and surrounding area, highlighting relevant regions, and reshaping user-drawn shapes to align with the ground plane elements, thereby simplifying rule creation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If users manually define regions by drawing shapes on video frames, then users can create custom monitoring zones, but the accuracy of region definition deteriorates when users lack understanding of ground plane geometry and occlusion effects
Solution Approach 1:
The system introduces an automatic region definition mechanism that acts as an intermediary between the user's simple rectangular selection and the complex ground plane geometry. The processor automatically calculates the intersection between the user-defined rectangle and the detected ground plane, occluding objects, and predicted pathways, thereby mediating the translation from simple user input to accurate monitoring regions without requiring users to understand complex geometric concepts
Solution Approach 2:
The system replaces the manual mechanical process of precisely drawing complex ground plane shapes with an automated computational system. Instead of users manually calculating and drawing accurate ground plane intersections, the processor automatically performs geometric calculations to determine the precise monitoring region based on the user's simple rectangular selection and the detected scene elements
2Ease of operation
If users draw tripwire lines in 2-D video space, then users can define crossing detection rules, but the effectiveness deteriorates when lines are drawn in incorrect locations such as sky areas where persons' heads might cross but feet never will
Solution Approach 1:
The system introduces an automated validation mechanism that acts as an intermediary between the user's 2-D line drawing and the 3-D ground plane evaluation. The processor projects the drawn line onto the detected ground plane and verifies whether the line intersects with predicted person pathways at appropriate heights, thereby mediating the translation from simple 2-D drawing to effective 3-D detection while preventing false placements in sky areas
Solution Approach 2:
The system performs preliminary analysis by automatically detecting the ground plane, occluding objects, and predicting person pathways before the user draws the tripwire line. This preliminary action provides the system with advance knowledge of valid detection locations, enabling it to validate and adjust user-drawn lines to ensure they are placed in effective positions for detecting actual person crossings rather than false sky-area crossings
3Manufacturing precision
If users manually adjust regions to account for occluding objects like trees and doors, then detection accuracy can be improved, but the time and complexity required for rule creation increases significantly
Solution Approach 1:
The system enables self-service by automatically detecting occluding objects such as trees and doors, and autonomously adjusting the monitoring regions to account for these occlusions. The processor performs automatic region refinement by identifying occluding objects and modifying the regions to exclude areas blocked by these objects, thereby eliminating the need for users to manually adjust regions while maintaining high detection accuracy
Solution Approach 2:
The system replaces the manual mechanical process of analyzing and adjusting regions for occlusions with an automated computer vision system. Instead of users visually inspecting and manually modifying regions to account for trees and doors, the processor automatically detects occluding objects through image analysis and performs the necessary geometric adjustments to the monitoring regions, thereby substituting complex manual operations with automated computational processes
4Productivity
If the system automatically creates regions based on scene analysis, then rule creation time is reduced, but the complexity of the system increases due to requirements for ground plane detection and scene understanding
Solution Approach 1:
The system achieves universality by designing a multi-functional processor that performs multiple tasks: detecting ground planes, identifying occluding objects, predicting person pathways, and automatically creating monitoring regions. This single processor handles diverse functions that would otherwise require separate systems, thereby enabling automatic region creation without proportionally increasing overall system complexity
Solution Approach 2:
The system performs preliminary scene analysis by automatically detecting ground planes, occluding objects, and predicting pathways before the user needs to create any rules. This preliminary action prepares the system with pre-computed scene understanding data, enabling rapid automatic region creation when users initiate rule creation, thereby achieving high productivity without requiring complex real-time processing during the actual rule creation moment
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Techniques are described for assisted creation of video rules via scene analysis. In some implementations, a scene is obtained, a shape of an element in a ground plane shown in the images of the scene is identified, user input that defines a shape of a region of interest used in a video rule is obtained, that the shape of the region satisfies a reshape criteria based on the shape of the element in the ground plane shown in the images of the scene is determined, and a reshaped region based on the shape of the element in the ground plane shown in the images of the scene is determined.