Introducing the Metis/DigitalGlobe Data Challenge

Last year, I completed the Metis Data Science Bootcamp in New York. I finished hiking the entire Appalachian Trail at the end of 2015, then spent 6 grueling months looking for a job. Once I accumulated a few hundred rejection letters, and it seemed like I wasn’t any closer to a job, I realized my skills …

What is Tomnod?

Out in the real world, one of the best ways to find things or people that are lost, whether lost keys or lost kids, is to recruit help to find them. In a nutshell, that’s what Tomnod is. In fact, it has been called a “virtual search party”. Tomnod is a crowdsourcing application that lets …

Push and Pull

In the previous post, we’ve discussed the challenges of scaling object detection. We talked about how the traditional model of tiling images doesn’t quite fit because we need to be able to detect objects that span multiple tiles. In this post, we’ll examine the challenge of implementing a system that will do that efficiently. We …

Measuring Performance for Object Detectors – Part 2

This post is the second and final installment in my series about measuring the accuracy of models. In my last post, I talked about how to choose a method to score models. Now it’s time to use that method to score models against each other. Let’s start with a brief summary of the goal. We …

Frontness: Applications of Geospatial Object Detection

In my work on DeepCore, I have pleasure of being exposed to many of the applications that benefit from automated geospatial object detection. Machine learning is the key to unlocking data’s potential, and some of the applications are completely intractable without automated techniques. The DeepCore team has worked hard to develop software for these problems. …

Scaling Object Detection

Object detection in geospatial context presents some unique scalability challenges that are not normally tackled in the machine learning field. DeepCore attempts to address these challenges, though there’s still work to be done. In this post we will discuss these issues and show how DeepCore will address them in the future. In most applications, CNNs …

Creating Synthetic Clouds in Python

Tiny Changes Can Fool AI There has been much discussion more recently (and some not so recently) on how minute changes to images can fool the smartest neural nets. Sharif et al. showed how to fool a neural net into classifying a Reese Witherspoon photo as Russell Crowe by adding a groovy pair of technicolor …

Introduction to the OpenSpaceNet User Interface

In this video tutorial we’ll give a brief overview of OpenSpaceNet, our easy-to-use interface that enables you to automatically detect objects in satellite imagery, using the latest in machine-learning technology. If you have any questions, please feel free to Contact Us. A complete transcript of the video is available here. Thanks for checking out OpenSpaceNet, …

Generating and Applying a Bag of Visual Words Model for Image Classification

Last year in June of 2016, I was hired at DigitalGlobe to provide data science support to a team located in Herndon, Virginia. The majority of my previous academic and work experience has been in the fields of image processing, computer vision, and remote sensing. I was excited and eager to make a career-transition as …

Measuring Performance for Object Detectors – Part 1

This is the first part of a 2-part series of posts about measuring the accuracy of detector models. As we develop more models, it’s becoming important to have a standard score that tells us how well each model performed on a problem so we can choose the best one for each application. In order to …