Boundary-Map Object Segmentation for Occluded Image Instances
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object identification methods in images struggle with fragmented parts, particularly in complex environments with occlusions, varying body proportions, and clothing, leading to challenges in creating meaningful object and instance segmentations.
Innovation Solution
An apparatus and method using a neural network to predict a boundary map, combined with a post-processing engine to cluster and refine object instances, effectively handling complex images with heavy occlusions by dynamically adapting to input characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If edge-based segmentation detection and computer vision methods are used to identify objects in images, then object identification capability is improved, but performance deteriorates when objects appear as fragmented parts due to occlusions and varying body proportions
Solution Approach 1:
The patent applies segmentation by dividing the image processing task into multiple stages: initial object detection, fragmentation identification, and reassembly of fragmented parts. The system segments the complex task of identifying occluded objects into manageable steps, first detecting visible parts and then reconstructing complete object instances from fragmented detections across multiple image regions.
Solution Approach 2:
The patent transitions from two-dimensional spatial analysis to three-dimensional spatial reasoning by inferring the three-dimensional structure and position of occluded objects based on visible fragments. This dimensional extension allows the system to reconstruct complete objects by considering depth and spatial relationships beyond the immediate image plane.
2Productivity
If traditional object detection methods are applied to complex images with heavy occlusions, then processing speed is maintained, but detection quality and completeness deteriorate
Solution Approach 1:
The patent performs preliminary actions by pre-identifying potential object regions and fragmentation patterns before final object reconstruction. The system prepares by detecting edge fragments and potential object boundaries in advance, organizing this preliminary information into structures that facilitate rapid final assembly and identification, thus maintaining processing speed while improving detection quality.
Solution Approach 2:
The patent introduces intermediary data structures and processing layers that bridge initial fragment detection and final object identification. These intermediaries include fragment clustering structures, boundary map representations, and instance segmentation data that mediate between raw detection data and final object conclusions, enabling both speed and accuracy.
Data Source
AI summary
An apparatus is provided. The apparatus includes a communications interface to receive raw data from an external source. The raw data includes a representation of a first object and a second object. The apparatus further includes a memory storage unit to store the raw data. In addition, the apparatus includes a neural network engine to receive the raw data. The neural network engine is to generate a segmentation map and a boundary map. The apparatus also includes a post-processing engine to identify the first object and the second object based on the segmentation map and the boundary map.