Individual Plant Localization Using Image-Attribute Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies are unable to accurately distinguish between individual plants in a field, which is essential for tracking growth, disease progression, and managing agricultural resources effectively, due to insufficient accuracy of position coordinates from sensors like GPS and IMU, especially when plants are close together.
Innovation Solution
Training machine learning models, such as convolutional neural networks (CNNs) and sequence-to-sequence models, using digital images and additional plant attributes like position coordinates, bounding shapes, and temporal data to recognize individual plants by generating unique identifiers or associations between images, allowing for more precise identification and tracking over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If position coordinate sensors (GPS, IMU) are used to identify individual plants, then the system can obtain position information, but the measurement precision is insufficient to distinguish between closely spaced plants
Solution Approach 1:
The patent introduces an intermediary coordinate transformation system that converts GPS/IMU coordinates into a local field coordinate system. This intermediary layer processes the imprecise global coordinates through mathematical transformation and integration with image data to achieve precise plant-level positioning without requiring higher-precision sensors
Solution Approach 2:
The patent merges multiple data sources including position coordinates, image data, and plant attributes into a unified plant identification system. By combining these diverse data types and processing them together through coordinate transformation and matching algorithms, the system achieves precision beyond what any single sensor could provide
2Measurement precision
If machine learning models are trained with multiple plant attributes to recognize individual plants, then the recognition accuracy improves, but the device complexity and processing requirements increase
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning models with multiple plant attributes (position coordinates, bounding shapes, temporal data) before deployment. This pre-training phase prepares the models to recognize individual plants accurately, reducing the need for complex real-time processing during actual plant identification
Solution Approach 2:
The patent utilizes parameter changes by varying multiple plant attributes (position, shape, temporal characteristics) as input parameters to the machine learning models. By changing and analyzing multiple parameters simultaneously, the system achieves high recognition accuracy while managing complexity through systematic parameter processing
3Measurement precision
If multiple plant attributes (position coordinates, bounding shapes, temporal data) are collected and processed, then the ability to distinguish individual plants improves, but the loss of time for data processing increases
Solution Approach 1:
The patent collects and processes multiple plant attributes in advance during data collection phases. By preparing position coordinates, bounding shapes, and temporal data beforehand and storing them in databases, the system reduces real-time processing requirements when individual plants need to be identified
Solution Approach 2:
The patent creates copies of plant data including position coordinates, image data, and attributes that are stored in databases. These pre-created data copies can be quickly retrieved and matched without重新 processing the original data, significantly reducing processing time during plant identification
Data Source
AI summary
Implementations are described herein for training and applying machine learning models to digital images capturing plants, and to other data indicative of attributes of individual plants captured in the digital images, to recognize individual plants in distinction from other individual plants. In various implementations, a digital image that captures a first plant of a plurality of plants may be applied, along with additional data indicative of an additional attribute of the first plant observed when the digital image was taken, as input across a machine learning model to generate output. Based on the output, an association may be stored in memory, e.g., of a database, between the digital image that captures the first plant and one or more previously-captured digital images of the first plant.