Human Detection in Overhead Images Using Score Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer vision systems face challenges in accurately detecting humans in top view images due to high perspective distortion and deformable body shapes, making it difficult to understand and analyze overhead tracking data effectively.
Innovation Solution
A system and method that assign scores to objects in images based on shape, motion, and location, tracking these objects across sequences of images, and determining human presence by comparing scores to thresholds, which are inversely related to initial scores, to enhance the accuracy and efficiency of human detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If overhead tracking cameras are used to monitor a space, then the system can provide comprehensive coverage of the area, but the images suffer from high perspective distortion and deformable body shapes making human detection difficult
Solution Approach 1:
The system performs preliminary actions by detecting objects in the first image, assigning an initial score based on shape, motion, and location parameters, and tracking the object across frames before making the final human detection determination. This preliminary scoring and tracking enables the system to handle perspective distortion by accumulating evidence across multiple frames rather than relying on a single distorted view
Solution Approach 2:
The system uses feedback by comparing the initial score from the first image with a threshold that is inversely related to the initial score when determining human presence in the second image. This feedback mechanism allows the system to adjust its detection sensitivity based on the quality of initial detection, improving overall accuracy despite perspective distortion in overhead views
2Difficulty of detecting and measuring
If shape detection algorithms are applied to top view images to detect human shapes, then the system can identify objects in the space, but the high perspective distortion and deformable body shapes reduce detection reliability
Solution Approach 1:
The system performs preliminary shape detection and scoring in the first image before tracking to the second image. By establishing an initial detection score based on shape, motion, and location parameters, the system creates a foundation for more reliable final determination that compensates for the unreliability of single-frame shape detection in distorted overhead views
Solution Approach 2:
The system maintains continuity of useful action by tracking the detected object across multiple frames and continuously updating the detection determination. This continuous tracking approach ensures that detection reliability is maintained over time despite perspective distortion, as the system accumulates consistent detection evidence across the image sequence
3Measurement precision
If the system uses score-based determination with threshold comparison to improve detection accuracy, then human detection precision improves, but the system complexity increases due to multiple scoring and threshold calculation steps
Solution Approach 1:
The system applies parameter changes by using a threshold that is inversely related to the initial score rather than a fixed threshold. This dynamic parameter adjustment optimizes detection accuracy by adapting the decision criterion based on the quality of the initial detection, improving precision without requiring extremely complex multi-parameter models
Data Source
AI summary
Embodiments of the invention relate to a method for determining that an object in a sequence of images is a human. The method may include the steps of detecting an object in a first image from the sequence of images and assigning a first score to the object. The object is tracked to a second image from the sequence of images and a second score is assigned to the object in the second image. The second score is compared to a threshold that is inversely related to the first score and a determination that the object in the second image is a human is made based on the comparison.