Object Detection via Probability Hypothesis Density Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection technologies fail to accurately distinguish between overlapping objects in image data, often merging multiple overlapping objects into a single detection.
Innovation Solution
An information processing apparatus that generates likelihood data for partial regions within image data, computes a probability hypothesis density (PHD) to represent the likelihood of object existence, and extracts partial distributions to distinctly identify overlapping objects by determining their positions and sizes based on statistical analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-maximum suppression is used to eliminate erroneous outputs, then detection accuracy for non-overlapping objects is improved, but overlapping objects are erroneously merged into a single detection
Solution Approach 1:
The patent segments the detection process by introducing an extraction unit that divides the probability distribution map into multiple partial distribution maps, each corresponding to a distinct object. This segmentation allows overlapping objects to be separated into independent detection results, resolving the contradiction where non-maximum suppression merges overlapping objects while maintaining detection accuracy for non-overlapping cases.
2Productivity
If region proposals are reduced by non-maximum suppression, then processing efficiency is improved, but cases of significant overlap are not considered and multiple objects are missed
Solution Approach 1:
By segmenting the probability distribution into partial distributions before output, the system maintains processing efficiency while improving detection completeness. Each partial distribution represents a distinct object, ensuring that overlapping objects are not merged and all objects are detected, thus resolving the contradiction between efficiency and completeness.
Solution Approach 2:
The extraction unit acts as an intermediary between the generation unit and output unit, processing the probability distribution map to extract distinct partial distributions. This intermediary step ensures that overlapping objects are properly separated while maintaining processing efficiency, resolving the contradiction between productivity and measurement precision.
Data Source
AI summary
An information processing apparatus (2000) generates likelihood data for each of a plurality of partial regions (12) in image data (10). The likelihood data are data being associated with a position and a size on the image data (10) and indicating a likelihood that a target object exists in an image region at the position with the size. The information processing apparatus (2000) computes a distribution (probability hypothesis density: PHD) of an existence likelihood of a target object with respect to a position and a size by computing the total sum of likelihood data each piece of which is generated for each partial region (12). The information processing apparatus (2000) extracts, from the PHD, partial distributions each of which relates to one target object. For each extracted partial distribution, the information processing apparatus (2000) outputs a position and a size of a target object represented by the partial distribution, based on a statistic of the partial distribution.