Computer Vision Yield Mapping for Horticultural Harvesting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current horticultural yield monitoring technologies face challenges in correlating yield to specific picking areas within a field due to limitations in existing technologies and the physical characteristics of horticultural crops, making it difficult to achieve accurate yield mapping during harvesting.
Innovation Solution
The implementation of a computer vision-based system that uses video imagery captured during harvesting to detect and count horticultural products, annotating the data with geographic location and timestamp information, allowing for the generation of yield maps that can be used to optimize harvesting routes and operational parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional yield monitoring technologies are used, then the system is simpler and easier to operate, but the measurement precision of yield correlation to picking area is insufficient
Solution Approach 1:
The patent replaces traditional mechanical yield monitoring systems with a computer vision-based system that uses cameras, image processing algorithms, and machine learning models to detect, track, and count horticultural products. This substitution enables precise yield mapping by automatically associating product detection data with geographic location information, resolving the contradiction between measurement precision and system complexity.
Solution Approach 2:
The patent introduces an intermediary data processing layer that combines multiple data sources (camera imagery, GPS location, product detection algorithms) to create a unified yield map. This intermediary processing layer integrates information from different sources and formats, enabling precise yield correlation to picking areas while managing system complexity through structured data flow and processing pipelines.
2Measurement precision
If computer vision-based systems are implemented, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent segments the computer vision system into distinct functional modules: camera subsystem, image processing subsystem, product detection subsystem, tracking subsystem, and mapping subsystem. Each module performs a specific function and can be independently optimized or replaced, managing overall system complexity while maintaining high measurement precision through specialized processing at each stage.
Solution Approach 2:
The patent employs preliminary actions by pre-processing camera imagery to enhance product visibility, pre-training machine learning models to recognize specific horticultural products, and pre-establishing geographic coordinate systems before actual yield monitoring operations. These preliminary preparations reduce real-time processing complexity and improve detection accuracy during field operations.
3Productivity
If real-time yield mapping is performed, then productivity improves, but loss of time in data processing increases
Solution Approach 1:
The patent implements continuous yield mapping by performing product detection, tracking, and geographic mapping operations throughout the entire harvesting process without interruption. The system continuously processes camera imagery and updates yield maps in real-time, enabling immediate identification of high-yield and low-yield areas while maintaining uninterrupted harvesting operations, thus improving productivity without significant time loss.
Solution Approach 2:
The patent employs periodic processing cycles where the system rapidly processes batches of camera imagery at regular intervals during harvesting operations. This periodic action allows the system to maintain real-time awareness of yield distribution while managing data processing load through structured batching, enabling productivity improvements without excessive time consumption in data analysis.
Data Source
AI summary
Embodiments of the disclosed technologies are capable of inputting, to a machine-learned model that has been trained to recognize a horticultural product in digital imagery, digital video data comprising frames that represent a view of the horticultural product in belt-assisted transit from a picking area of a field to a harvester bin; outputting, by the machine-learned model, annotated video data; using the annotated video data, computing quantitative data comprising particular counts of the individual instances of the horticultural product associated with particular timestamp data; using the timestamp data, mapping the quantitative data to geographic location data to produce a digital yield map; causing display of the digital yield map on a field manager computing device.