Vision-Based Plant Detection with 3D AR Field Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing crop scouting technologies lack efficient methods for accurately detecting and analyzing plant conditions in three-dimensional agricultural spaces, leading to inaccuracies in weed infestations, nitrogen status, leaf damage, disease recognition, and pest monitoring.
Innovation Solution
A computer-implemented method using machine learning models, augmented reality, and computer vision to detect and plot plants in a three-dimensional space, incorporating convolutional neural networks and Fuzzy Logic rules to correct for device movement and improve plant classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional hand-based scouting methods are used, then operational simplicity is maintained, but measurement precision and detection accuracy deteriorate
Solution Approach 1:
The patent replaces manual mechanical scouting with an automated vision-based system using cameras, machine learning models, and computer vision algorithms to detect and analyze plant conditions, eliminating the need for manual measurement and observation
Solution Approach 2:
The system enables self-service scouting where the automated vision system independently performs plant detection, classification, and condition analysis without requiring skilled human scouts, allowing any user to obtain accurate agricultural data
2Measurement precision
If two-dimensional image analysis is used, then processing simplicity is maintained, but measurement precision of spatial relationships deteriorates
Solution Approach 1:
The patent transitions from two-dimensional image analysis to three-dimensional spatial understanding by projecting detected objects onto a ground plane and creating 3D representations of plants and field features, enabling accurate spatial positioning and distance measurement
Solution Approach 2:
The system introduces a ground plane projection as an intermediary layer that bridges 2D camera images and 3D spatial coordinates, allowing accurate mapping of detected objects to real-world positions without requiring complex direct 3D capture systems
3Productivity
If manual scouting methods are used, then device complexity is low, but productivity and scouting efficiency deteriorate
Solution Approach 1:
The vision-based scouting system operates continuously as the mobile device moves through the field, constantly capturing and analyzing images without interruption, whereas manual scouting is intermittent and depends on human movement and observation cycles
Solution Approach 2:
The patent replaces manual mechanical scouting operations with automated computer vision algorithms and machine learning models that rapidly process images to detect plants, weeds, and field conditions, dramatically increasing scouting throughput
4Reliability
If basic image recognition is used, then processing simplicity is maintained, but reliability of plant classification deteriorates
Solution Approach 1:
The system incorporates feedback mechanisms where machine learning models continuously refine their classifications based on confidence scores and can request re-evaluation or adjustment, improving classification reliability through iterative verification
Solution Approach 2:
The patent uses multiple machine learning models with different complexity levels and parameter settings that can be dynamically adjusted based on the specific detection task, plant type, and environmental conditions to optimize classification accuracy for different scenarios
Data Source
AI summary
In one embodiment, a method includes in response to an input to initiate a continuous process for scouting, obtaining image data from one or more sensors of a device, analyzing one or more input images from the image data, generating a grid of two dimensional (2D) reference points that are projected onto a ground plane to create a matching set of three dimensional (3D) anchor points, and their positions in a 3D space of the agricultural field using augmented reality (AR), providing the one or more input images, tracking grid, and the positions in the 3D space of the agricultural field to a machine learning (ML) model having a convolutional neural network (CNN), and generating inference results with the ML model including an array of detected objects and selecting most likely inferred plant locations and classifications for the array of detected objects.