The speed and accuracy of geospatial information define the quality of agronomic decisions in modern agriculture. Field boundary mapping, obstacle detection, and area zoning are fundamental for generating prescription maps — especially when planning drone-based spraying or variable-rate application of fertilizers and crop protection products.
Our Swiss partner,Pix4D, has released the PIX4Dfields 2.10 update, which introduces:
– The AI Object Selection feature for automated field boundary detection, obstacle identification, and area annotation
– An enhanced Magic Tool for precise weed detection in row crops
– A new data analysis mode for inspecting individual pixel values across RGB, DSM, and index layers
These tools significantly increase data processing speed and provide farmers with access to critical information without delays or manual annotation.
AI Object Selection: automated handling of boundaries, obstacles, and treatment zones
The central innovation in PIX4Dfields 2.10 is the AI Object Selection tool — an interface that enables users to automatically define field boundaries, detect obstacles, and create annotations within specific zones in just a few clicks.
This tool includes three operating modes:
- AI Boundary — automatic detection and outlining of field boundaries based on drone imagery. The software accurately traces field contours without manual input.
- AI Obstacle — detection of obstacles such as utility poles, water bodies, unplantable areas, or technical infrastructure. Selection is performed automatically, with the option for manual adjustment.
AI Obstacle та AI Boundary у PIX4Dfields
- AI Area Mode — a mode within the annotation toolbar that allows for rapid marking of localized zones: areas for targeted application, trees, or damaged crops, including post-flooding. This simplifies analytics and the creation of prescription maps for variable-rate actions.
AI Area annotation used for tree detection in PIX4Dfields
With AI Object Selection, annotating complex objects can be up to 10 times faster than manual marking. This not only accelerates the workflow but also reduces the risk of errors and delivers accurate, ready-to-use geospatial data.
Tree detection indicating also total tree area covered on the field (left) and automatic selection of damaged field areas by flood using AI Area annotation in PIX4Dfields
The feature is especially valuable for operators preparing aerial spraying missions: fast generation of boundaries and no-spray zones reduces preparation time and increases effective flight time, minimizing delays during the mapping stage. Additionally, the ability to precisely mark even small obstacles enhances operational safety, creating a clear and reliable foundation for planning.
Updated Magic Tool: precise weed detection in row crops
The Magic Tool, already proven as an effective solution for weed identification, has been significantly improved in version 2.10. The new algorithm enables highly accurate weed detection, even in challenging Green-on-Green scenarios where crops and weeds share similar spectral signatures.
The algorithm undergoes additional training based on imagery from row crops, allowing it to accurately distinguish weed vegetation even at early growth stages.
The Magic Tool detecting weeds in cabbage crop automatically
The Magic Tool detecting weeds on an onion field
Thistles in turnips detected by the Magic Tool (left) and weeds detected in potato row crop (right)
Weed detection is performed automatically, without the need for manual processing or zoning, significantly speeding up preparation for selective herbicide application. The Magic Tool analyzes the row structure and segments weeds accordingly.
This update expands the use of PIX4Dfields for planning zone-based spraying based on actual field conditions — especially relevant for farms practicing sustainable agriculture and aiming to optimize their CPP (Crop Protection Product) usage.
Point-based data analysis: numerical pixel values for RGB, DSM, and index layers
PIX4Dfields 2.10 introduces a new local data preview feature: users can click on any point on the map and retrieve precise numerical values for the corresponding pixel across all analytical layers in the project — including RGB imagery, the Digital Surface Model (DSM), and vegetation indices.
This tool enables users to:
- perform point-specific diagnostics of crop condition, terrain, or moisture directly at a selected coordinate;
- create custom indices tailored to specific crops or current agro-technological conditions;
- correlate drone data with field observations and validate the resulting analytics.
This functionality is especially valuable for agronomists working with heterogeneous fields, trial plots, stress zones, or conducting comparative analysis of treatment efficacy. Access to baseline values at a specific point enhances the accuracy of agronomic conclusions and supports objective interpretation of data obtained through remote sensing.
Getting individual pixel insights on trial plots in PIX4Dfields
Pixel value that can be used for RGB, DSM or index layers in PIX4Dfields
The PIX4Dfields 2.10 update is designed to enhance productivity and precision when working with geospatial data in agricultural production. AI-powered tools for automatic detection of boundaries, obstacles, and treatment zones, an improved algorithm for weed detection in row crops, and the ability to perform point-level layer analysis all help agronomists, UAV operators, and agricultural consultants reduce data processing time and improve decision-making quality in the field.
Read more about the solution update here:
https://www.pix4d.com/blog/pix4dfields-AI-object-selection
Purchase a licence for Pix4D products:
https://store.drone.ua/pix4d/
For consultation and license purchases, call:
063 526 04 54
