Crowdsourcing provides flexibility and innovation at scale. One of the most interesting uses of Topcoder’s crowdsourcing model is the ability to rapidly develop applications simultaneously. Over the past couple of years, organizations such as HPE and the U.S. Department of Energy (DOE) have used Topcoder to embark on Innovation Programs — each focused on a particular theme.
You might assume that programs like HPE’s Living Progress Challenge or the DOE’s SunShot Catalyst are only useful for developing simple mobile or web application prototypes. But in the context of these programs, we’ve been helping clients deal with significant data challenges — either with messy or unstructured data. At Topcoder, our goal is to get a working system up and running rapidly and begin giving insights to users within weeks, rather than months.
Here is a look at some of the data-driven applications Topcoder has delivered over the past year or so as part of our Innovation Program efforts.
Increased efficiency through an object tracking application
Samasource had a problem. Their agents were spending a lot of time tagging objects in video footage provided by their clients. Oftentimes, agents found the same (or very similar) objects in frame after frame. Over a period of 12 weeks in the Living Progress Challenge, Topcoder developed an application using Python and OpenCV to track objects in video footage. We provided tagged objects in a set of images once every 15 frames and our application identified and tagged the same (untagged) objects in subsequent frames.
It’s much faster and easier to make a minor adjustment to an already tagged object than to insert all of the tags in a particular image — starting from the ground up. This type of functionality can dramatically improve the efficiency of Samasource’s agents and allow them to stay competitive.
Trade data comparison and the fight to end illegal fishing
The World Wildlife Fund’s oceans team wanted to develop an application to compare UN Comtrade trade data reported by trading partners to find inconsistencies that might suggest illegal activity. Using HPE’s Vertica database, a set of C# REST-based services, and a web front-end using AngularJS, the Topcoder Community produced an analytics and communication tool to combat the illegal fish trade.
Automated classification of civic improvement suggestions
Stanford researchers experimenting with crowdsourced democracy and civic engagement faced a daunting task: analyzing a complex, unstructured data set and providing guidance to civic leaders about the concerns of their citizens. Topcoder developed both Haven OnDemand and Python-based supervised learning algorithms to automate the classification of comments from public-facing websites and even suggested visualization ideas to facilitate analysis. Dr. Tanja Aitamurto described the process in greater detail in her article “Civic CrowdAnalytics: making sense of crowdsourced civic input with big data tools.”
Streamlined categorization of water source data
The Water Point Data Exchange is the global platform for sharing water point data. And researchers and policymakers need to understand the current state of water access in order to make recommendations and coordinate development of safe drinkable water — a critical and scarce resource in the developing world. Of course, manually categorizing water source records provided by field personnel is a difficult and time-consuming task. With the Living Progress Challenge, Topcoder developed Python-based supervised learning algorithms to enable water source-based classification of records and technology to support the analysis of this critical data.
The Living Progress Challenge is just one example of some of the work Topcoder has done with Innovation Programs. (Recently, it also won HPE the ISPIM Grand Prize 2017 for Excellence in Innovation Management.) It involved many technical challenges and provided opportunities for members of the Topcoder Community to experiment and execute on ideas quickly for different organizations.
Learn more about our Innovation Programs and let us help you start on your next big idea.