2D Food Quantity Measurement Using Learned Region and Depth Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current food quantity measurement technologies using 2D images suffer from low accuracy due to insufficient differentiation between food types and quantities, especially when not using 3D scanners, leading to errors in calorie counting and food management.
Innovation Solution
A method and apparatus that learn an object region model and object quantity model using 2D images, employing feature maps and regression loss to accurately segment objects and measure their quantities, without requiring depth cameras or 3D scanners.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If 2D images are used for food quantity measurement, then device complexity and cost are reduced, but measurement precision deteriorates
Solution Approach 1:
The patent applies dimensionality change by introducing depth estimation through monocular depth prediction algorithms. The system processes 2D images and generates depth maps that estimate the third dimension (depth), effectively transforming 2D image data into 3D spatial information. This allows accurate volume calculation of food items using only 2D camera inputs, resolving the contradiction between using simple 2D devices and achieving precise measurements.
Solution Approach 2:
The patent introduces an intermediary depth map as a mediator between the 2D image and the final volume measurement. The depth map serves as an intermediate representation that bridges the gap between 2D image data and 3D volume calculation, enabling accurate measurement without requiring direct 3D scanning hardware.
2Measurement precision
If 3D scanners are used for food quantity measurement, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent creates a virtual 3D copy of the food item through depth map generation and 3D reconstruction algorithms. Instead of using physical 3D scanning hardware, the system generates a digital depth representation from 2D images, which is then used for volume calculation. This copying approach achieves 3D measurement precision without the complexity of physical 3D scanners.
Solution Approach 2:
The patent replaces the mechanical 3D scanning system with a computational approach using monocular depth prediction. Instead of using mechanical sensors and multi-camera systems, the invention uses AI-based algorithms to estimate depth from single 2D images, substituting mechanical complexity with computational processing.
3Ease of operation
If simple image search is used for food identification, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent changes the parameters used for food identification from simple image matching to multi-parameter analysis including color histograms, texture features, and depth information. By incorporating multiple parameters beyond basic image search, the system achieves both ease of operation (automatic identification) and high measurement precision (accurate food type recognition for proper calorie calculation).
Data Source
AI summary
Provided are a method and an apparatus for measuring an object quantity using a 2D image. A method for measuring an object quantity using a 2D image according to one embodiment of the present disclosure comprises learning an object region model in a first object image which is a pre-learning target, learning an object quantity model for a first object region in the first object image using a feature map extracted from the first object image, and measuring an object quantity of at least one of a second object region and a background region in a second object image which is a measurement target using the learned object region model and object quantity model.