From spotting damaged structures post-natural disaster to classifying land use change to identifying tree species based on a picture of the leaf, our work in machine learning and computer vision can help you create iterative processes that enable decision-making, fuel advanced studies, and drive positive change.
We’ll help you define both the problem and potential solutions. Our experienced team will recommend the process that fits your need, whether it be semantic segmentation, classification, regression, or object detection.
We’ll define a labeling process and engineer or compile the tools you need to get the job done. Our partners can help annotate data to validate and create training data sets.
Our team will write custom scripts or leverage our Raster Foundry API to create a data set in a friendly format that works with other tools.
Training and evaluating models
We’ll evaluate architectures and parameters to construct a performant model that identifies features, classifies imagery, or detects objects across huge data sets.
By engineering a service to make predictions that fit your specific use-case, we’ll help you make calculated decisions.
Designing custom apps
We offer software engineering, user experience design, and data analytics services - our well-rounded team can help you construct a custom user interface to fit your process. Check out our open source Raster Foundry app as an example!
One of our mature products, HunchLab, utilizes machine learning techniques to create a unified forecast of crime risk for sophisticated prevention analysis. Advanced statistical models forecast when and where crimes are likely to emerge.
OpenTreeID, a research project with the USDA, is aimed at advancing community forestry with human-augmented computer vision. The mobile site uses a deep learning model trained on a database of images to recognize the species of leaves.
Check out our blog
We write about our methods, projects, and research related to geospatial machine learning.