Plant Element Characteristic Determination Using Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining the protein content and characteristics of plant elements, such as wheat grains, are costly and require samples to be sent to specialized laboratories, limiting their accessibility and efficiency, especially for on-site analysis in the agricultural and food industries.
Innovation Solution
A method utilizing image analysis software and an automatic learning machine to classify and determine characteristics of plant elements by training on image matrices of isolated plant elements, allowing for the calculation of characteristic values and indicators of inclusion in classes, which can be implemented using convolutional neural networks or other algorithms, reducing the need for extensive sample analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods (Kjeldahl, Dumas, NIRS) are used to determine protein content, then measurement precision is improved, but cost increases and requires sending samples to specialized laboratories
Solution Approach 1:
The patent uses digital images as copies of physical plant elements to create training data for machine learning models. Instead of analyzing physical samples through expensive laboratory methods, the system captures images and uses them to train automatic learning machines, replacing costly physical analysis with computational analysis of image copies.
Solution Approach 2:
The patent replaces mechanical and chemical analysis systems (Kjeldahl, Dumas methods) with an information-based system using image processing and machine learning. The physical-chemical measurement process is substituted by capturing visual information and processing it through automatic learning machines, eliminating the need for complex laboratory equipment and procedures.
2Reliability
If traditional laboratory methods are used, then reliable results are obtained, but loss of time increases due to sample transportation and processing delays
Solution Approach 1:
The patent performs preliminary action by capturing images of plant elements at the source location before any analysis is needed. These images are stored and can be processed immediately or at any later time without requiring physical sample preservation or transportation, enabling analysis to be performed at the optimal moment while preserving all visual information.
Solution Approach 2:
The patent introduces digital images as an intermediary between the physical plant elements and the analysis process. Instead of transporting physical samples through the supply chain to laboratories, images serve as intermediaries that can be transmitted, stored, and analyzed remotely, decoupling the timing and location of sampling from analysis.
3Productivity
If image analysis is used to determine plant element characteristics, then cost and time are reduced, but measurement precision may be compromised
Solution Approach 1:
The patent performs preliminary training of automatic learning machines using large datasets of images paired with reference measurements from reliable methods. This preliminary action creates pre-trained models that have already learned to associate visual features with accurate characteristic values, enabling fast inference without sacrificing precision during actual analysis.
Solution Approach 2:
The patent implements feedback mechanisms where the machine learning model's predictions are continuously refined using feedback from reference measurements. The system learns from discrepancies between predicted and actual values, adjusting its parameters to improve precision while maintaining the speed and cost advantages of image-based analysis.
Data Source
AI summary
A method for training an automatic learning machine (3.12), comprising the following steps:defining, in a database, at least one first class of image matrices (3.11) and one second class of image matrices (3.11), the first class of image matrices being associated with said first measured characteristic value;the second class of image matrices (3.11) being associated with said second measured characteristic value;classifying said image matrices in said classes as a function of the respective measured characteristic value for the mixture of plant elements from which each of said images has been obtained;modifying, by said calculator (3.15), in response to the reception of said overall classification error (3.18), at least one of its calculation elements, so that the second probabilities of inclusion of the image matrices in said at least two classes are closer to the true inclusion of the at least three image matrices in the at least two classes than the first inclusion probabilities.