September 17, 2024
The Global Mapper Insight and Learning Engine™(beta) Leads Expansion of Analysis Tools in Global Mapper Pro v26.0
The fall update to Global Mapper Pro, v26.0, is headlined by the cutting-edge Global Mapper Insight and Learning Engine™(beta), a deep learning-powered image analysis toolset. This new version also introduces a brand new Solar Analysis tool designed to analyze 3D data and calculate regions of sun/shadow over time. Global Mapper Pro’s machine learning-based Automatic Point Cloud Analysis tools, which feature custom training and classification, contain a variety of improvements as well. Photogrammetric analysis continues to improve with the addition of a workflow and speed updates to the Pixels to Points tool.
As always, the complete list of new features, updates, and bug fixes is included in the v26.0 Knowledge Base. Remember that if your maintenance and support subscription is active, you can upgrade your Global Mapper to version 26.0 at no cost. If you have questions about your order status, please contact our licensing team at authorize@bluemarblegeo.com.
Keep an eye out for the next blog: Improved Ease-of-Use, Editing, and Data Support in Global Mapper v26.0!
Global Mapper Insight and Learning Engine™(beta)
The Global Mapper Insight and Learning Engine™(beta) opens the door for deep learning-based image analysis in Global Mapper. All Global Mapper Pro v26 users have access to this functionality through the Deep Learning toolbar and menu, which boasts a variety of automated raster analysis tools. This new suite of tools provides trained models for land cover classification, vehicle identification, and building extraction. A fine-tuning option allows users to re-train layers of a model to improve the analysis results on specific datasets.
Any user with a valid Global Mapper Pro v26.0 license can access the Global Mapper Insight and Learning Engine tools until the next version release (typically in late February). Don’t worry, the deep-learning tools will still be available in the next version of Global Mapper, you’ll just need to update to the new version to continue using them and see what else we have been working on!
Land Cover Classification
The Land Cover Classification (LCC) process in Global Mapper gives users the ability to conduct LCC analysis themselves. The tool is designed to work with higher resolution source data when compared to existing publicly available datasets, which are often done at a regional or global scale. This provides users with the ability to create higher accuracy and higher resolution products for their specific areas of interest.
Object Detection
Object detection refers to the task of identifying discrete objects in an image. The Global Mapper Insight and Learning Engine currently provides models for two types of object detection: vehicles and buildings.
Trained on high-resolution aerial and satellite imagery, vehicle detection methods identify vehicles in an image via a vector bounding box. Building identification — often referred to as building extraction since it identifies explicit building boundaries — produces both vector building bounds and a binary raster mask.
After
Middle
Before
Fine-Tuning and Training
Arguably the most powerful component of these new tools is the ability to fine-tune and retrain the built-in image analysis models. The Global Mapper Insight and Learning Engine (beta) is designed for fine-tuning and retraining models. This functionality enables users to tailor models to specific datasets and tasks. Users can import, or manually create ground truth data tied to their sample imagery and run the fine-tuning process. This creates a custom model, augmented based on this ground truth data, for future use in the application. Fine-tuning provides the user with control over how deeply the model is adjusted, where full training adjusts the entire model.
Solar Analysis
Shadow percent and coverage area can now be calculated in Global Mapper Pro based on analyzing terrain, 3D vectors, and mesh features. Understanding the impact that terrain and surface features can have on the landscape applies to a variety of scenarios — from agriculture and urban planning to commercial and residential solar power applications. This new tool allows users to calculate shadow masks and shadow coverage percentage for a given area. The tool can account for a time range, height above ground, and create an animation layer as the output to be used in the Animate Layers tool.
Automatic Point Cloud Analysis
This release includes numerous updates to the Automatic Point Cloud Analysis tool. This cutting-edge tool available in Global Mapper Pro was recently recognized at the 2024 Lidar Leader Awards for winning the Outstanding Innovation in Lidar Award. The updates in version 26.0 continue to establish Global Mapper Pro as the go-to solution for lidar and point cloud workflows.
Model Key Point Identification
The identification of model key points is an important process used to reduce the size of a data set, both in point count and file size, while still retaining important structural qualities of the data. Model key points are often used when subsequent analysis is required outside of a traditional geospatial application, such as CAD environments or gaming and simulation scenarios.
Added as a new section to the Automatic Point Cloud Analysis tool, Model Key Points allows users to identify these points either as a Flag or Classification, per ASPRS guidelines. Users can optionally create a new layer of these points, and/or a derivative TIN.
Saving/Sharing of Custom-Trained Classification Models
Version 25.0 of Global Mapper Pro introduced the training of custom point cloud classification models for use in the Automatic Point Cloud Analysis tool. This analysis, based on segmentation and the creation of a custom feature model, is improved in version 26.0 of the program with the option to save and load trained custom classes. These new options in the Automatic Point Cloud Analysis allow users to save a trained class out of Global Mapper Pro in a file that can be shared to another machine or Global Mapper user to be loaded into the tool.
Pixels to Points Updates
Pixels to Points is an invaluable tool for Global Mapper Pro users who work with drone-collected data. With the proliferation of drones from the professional to the consumer level, ease of image acquisition via drone flights, and value of derivative products, photogrammetric analysis is essential to many workflows and has a variety of applications for users working in the field. In version 26, Global Mapper Pro expands upon these powerful tools to include even more powerful automatic ground control point (GCP) placement and continued speed improvements.
Advancements to Automatic Ground Control Point Placement
Automatic GCP placement was an exciting new feature added to the Pixels to Points tool in version 25.1. Naturally, we wanted to continue to improve that process for our users in Global Mapper Pro v26.0. New updates to this tool further refine the identification process in scenarios of extreme terrain. Non-standard targets are now more easily identified and improved scoring allows for more accurate placement when multiple targets might be visible in one image.
Speed Improvements
Photogrammetric processing and analysis can be lengthy tasks because of their computationally-intense nature. We’ve further improved our processing pipeline, which benefits CPU and GPU-based processing. We’re proud to report that this version of the software boasts speed improvements of upwards of 50% in some scenarios compared to the previous version of the tool. We encourage anyone who has experience with photogrammetric processing to check out the speed improvements we’ve made in Global Mapper Pro v26.0.
To further explore deep learning image analysis and many other updates in version 26.0, download Global Mapper Pro and register for a 14-day free trial!
