Automated Crop Phenology Image Filtering via NDVI Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining phenotypic data in crops are often manual, resource-intensive, and inconsistent, lacking accuracy and efficiency in data collection and analysis.
Innovation Solution
The system processes image data from fields and plots using unmanned aerial vehicles (UAVs) to calculate normalized difference vegetation index (NDVI) values, flags inconsistent images, and trains a model to predict phenotypic data based on image and phenotypic data, reducing manual data collection and enhancing accuracy and consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual methods are used to determine phenotypic data in crops, then flexibility and adaptability are maintained, but efficiency and consistency deteriorate
Solution Approach 1:
The patent replaces manual mechanical assessment methods with automated image capture systems (UAVs, satellites) and computational analysis (NDVI calculations, machine learning models) to determine crop phenotypic data, thereby increasing efficiency and consistency while reducing manual labor
Solution Approach 2:
The system enables self-service by allowing the image processing and phenotypic data determination to occur automatically through computational algorithms and machine learning models without requiring continuous manual intervention, with the system processing images and generating predictions autonomously
2Measurement precision
If high-resolution images are captured frequently to improve phenotypic data accuracy, then measurement precision improves, but resource consumption and data processing complexity worsen
Solution Approach 1:
The patent extracts only the most relevant information from images by calculating NDVI values and using machine learning models to identify key phenotypic features, rather than processing all raw image data, thereby reducing processing complexity while maintaining accuracy
Solution Approach 2:
The system transforms raw image data into derived parameters (NDVI values) that capture essential phenotypic information in a condensed form, making the data more manageable and easier to process while preserving the accuracy needed for phenotypic assessment
3Reliability
If inconsistent images are included in the data set, then data quantity is maintained, but model training accuracy deteriorates
Solution Approach 1:
The patent implements feedback mechanisms where NDVI values from sequential images are compared to identify inconsistencies, and machine learning models evaluate image quality, providing feedback that triggers filtering or removal of problematic images before model training, thereby ensuring high reliability
Solution Approach 2:
The system performs preliminary filtering and validation of images through NDVI analysis and quality checks before images are used for model training, removing inconsistent or low-quality images in advance to prevent them from degrading model accuracy
Data Source
AI summary
Systems and methods are provided for use in processing image data of crops associated with one or more plots. One example computer-implemented method includes accessing a data set including images associated with one or more plots. The method then includes, for each plot, comparing a first index value of a first image of the plot at time n to an index value of a second image of the plot at time n+1; in response to the second index value being greater than the first index value, flagging the second image; and modifying the data set by removing at least part of the second image based on the flag. The method further includes accessing phenotypic data for the one or more plots at a time consistent with the images and training a model based on data including the modified data set and the accessed phenotypic data.