3D Product Counting From Single-Image Surface Coordinates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cycle counting methods using computer vision are limited by the field of view of the image capturing device, often requiring human intervention to account for products outside the view, leading to inefficiencies and errors.
Innovation Solution
A method utilizing a computing device with a recognition computer vision model to identify visible surfaces, generate image and base coordinates, associate surfaces with virtual products, and classify them as horizontal or vertical, enabling accurate product counting of both visible and obscured products based on a single image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If standard computer vision techniques are used for cycle counting, then the counting process is automated, but the field of view limitation causes products outside the view to be missed
Solution Approach 1:
The patent transitions from 2D image analysis to 3D spatial reasoning by generating a three-dimensional representation of the warehouse environment. This includes creating depth maps, identifying occlusion relationships, and modeling the three-dimensional positions of products, thereby capturing products that are hidden or outside the direct field of view in traditional 2D images.
Solution Approach 2:
The patent introduces an intermediary computational model that bridges the gap between limited camera views and complete product inventory. This model uses machine learning to predict the positions and visibility of products based on partial observations, occlusion patterns, and warehouse layout information, thereby inferring the presence of products not directly visible in the image.
2Measurement precision
If human-based counting is used to supplement computer-based count for products outside field of view, then counting accuracy is improved, but efficiency decreases and human errors are introduced
Solution Approach 1:
The system performs self-service by automatically detecting and counting products that are occluded or outside the field of view without requiring human intervention. The machine learning model independently infers the presence and position of hidden products by analyzing occlusion patterns, depth information, and spatial relationships, thereby maintaining both accuracy and efficiency without human supplementation.
Solution Approach 2:
The patent replaces the mechanical human counting process with an automated computational system. Instead of using human operators to physically locate and count products outside the field of view, the system uses computer vision algorithms, depth mapping, and machine learning models to automatically detect and count these products, thereby eliminating human errors while maintaining productivity.
3Measurement precision
If multiple images are captured to account for all products, then counting accuracy is improved, but time consumption and operational complexity increase
Solution Approach 1:
The patent performs preliminary action by capturing a single comprehensive image that is then enhanced through computational processing. Rather than requiring multiple sequential images, the system pre-processes the single image by generating depth maps, identifying occlusion zones, and using machine learning to predict hidden product positions, thereby achieving complete product detection from one capture moment.
Solution Approach 2:
The patent creates virtual copies and three-dimensional representations of the physical warehouse environment and products. By generating digital twins, depth maps, and spatial models from a single image, the system replicates the complete product inventory information without requiring multiple physical image captures, thereby reducing time while maintaining accuracy.
Data Source
AI summary
A product count of a number of physical products within a physical grouping of a plurality of the physical products is determined based on an image of the physical grouping. A cycle counter generates image coordinates for visible product surfaces within the image. The cycle counter generates three-dimensional virtual base coordinates for the expected locations of the plurality of the physical products within the physical grouping. The cycle counter determines the actual locations of the visible product surfaces within three-dimensional space based on a comparison of the image coordinates and the virtual base coordinates. The cycle counter determines the product count of the number of physical products within the physical grouping based on the actual locations.