Harvester Stalk Mass Estimation Using Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing agricultural harvesters face challenges in accurately monitoring the amount of processed crops during the harvesting operation, particularly in separating and quantifying stalks from debris.
Innovation Solution
A system utilizing a machine-learned mass estimation model that processes data from sensors and images to determine the mass of harvested material, including stalks and debris, by segmenting and calculating pixel proportions within a harvested material mask.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional monitoring methods are used for processed crops, then the system is simple, but the measurement precision of stalk mass is insufficient
Solution Approach 1:
The harvested material is segmented into distinct components (stalks and debris) using image processing and machine learning models. The system divides the monitoring task into separate analysis streams, processing different material types independently to achieve precise stalk mass measurement while maintaining manageable system complexity through modular architecture.
Solution Approach 2:
The patent replaces traditional mechanical or manual monitoring methods with optical sensing, image processing, and machine learning algorithms. This substitution enables non-contact, automated stalk mass measurement with high precision, transforming the monitoring system from simple but inaccurate to complex yet highly accurate.
2Productivity
If no real-time monitoring is implemented, then the device complexity is low, but the productivity and operational efficiency cannot be optimized
Solution Approach 1:
The system implements real-time feedback by continuously monitoring stalk mass and debris separation using sensors and image processing. This feedback loop provides operational data that can be used to optimize harvesting parameters, improve productivity, and maintain the harvester at optimal performance levels throughout the harvesting operation.
Solution Approach 2:
The monitoring system operates autonomously, using onboard sensors, processors, and machine learning models to self-evaluate the harvesting performance. The system automatically calculates stalk mass, separates debris quantification, and provides operational insights without requiring external intervention, thereby improving productivity while managing complexity through automation.
3Measurement precision
If debris and stalks are monitored together as one mass, then the measurement process is simple, but the measurement precision of separate components is lost
Solution Approach 1:
The system applies segmentation by separating the monitoring of stalks and debris into distinct measurement streams. Image processing algorithms and machine learning models identify and segment different material types based on visual characteristics, enabling precise quantification of each component separately while managing the complexity of differentiation through automated classification.
Solution Approach 2:
The system utilizes optical properties and visual characteristics (analogous to color changes) to differentiate between stalks and debris. Image sensors capture visual data that reveals distinct properties of different materials, allowing the machine learning models to accurately distinguish and quantify separate components despite the complexity of visual differentiation.
Data Source
AI summary
A system for an agricultural harvester includes one or more processors and one or more non-transitory computer-readable media that stores a machine-learned mass estimation model configured to receive data associated with one or more operation-related conditions for an agricultural harvester and process the data to determine an output indicative of a stalk mass for the agricultural harvester. The one or more non-transitory computer-readable media also stores instructions that, when executed by the one or more processors, configure the computing system to perform operations. The operations include obtaining the data associated with one or more operation-related conditions; inputting the data into the machine-learned mass estimation model; and receiving a value for the stalk mass as the output of the machine-learned mass estimation model for a defined time period.