Big Data Goes to the Birds: Fun at DeepScience
If you're like me and live in Evergreen, you never cease to be amazed by the variety of wildlife in this beautiful town. We're just 30 minutes from Denver but whenever the visible wildlife start to outnumber the people, you know you're in a different land.
One of my favorite recreational activities out here is enjoying nature - not too surprising. The other is working at my computer. Most people will think of those as two very different activities, just about as compatible as a family dinner where everyone is on their cell phones. I, however, see nature and computers as being highly compatible. With nature, you can have recreational activities. With computers, you can do research. With both combined, you can do research on how to conserve the recreational spaces we enjoy so much.
In Bergen Park on Saturday mornings, DeepScience hosts a small advanced study group around learning artificial intelligence. Besides following a set curriculum, we are also doing a project on bird identification. Imagine that, a big data/AI business studying birds. You would think there were more important things we could be doing, like finding treatments for type II diabetes or visualizing large datasets to find ways of lowering chemical exposure. (We do those too, but every now and then we just need to have a bit of fun.)
With some practice, most of us can look out the window and tell what bird is visiting a feeder. However, what if a bird has hit the window and flies off, or a bird recently becomes deceased and all that remains are a few clusters of feathers? It is relatively easy to write a mobile phone app that will identify birds telescopically, but no good app exists that can identify birds microscopically. Solving this problem would require pulling together computers, photography, programming cell phones, and some good old fashioned teamwork. I love solving problems that pull together multiple fields of science and technology to solve. That pretty much sums up what we do at DeepScience.
The first items our team would need were pictures of feathers. My search led me to the Evergreen Nature Center at the Evergreen Lake House next to downtown Evergreen. It is open from April to October and is a project of Evergreen Audubon, the local chapter of the National Audubon Society. If you haven't heard of Audubon, the name is synonymous with bird conservation. This is no small task. By protecting birds, we protect their habitats which protects our ecosystems which protects the natural world that keeps our society stable and ever marching forward. Folks who are members or volunteer at conservation organizations like Evergreen Audubon really are the unsung heroes of the natural world.
Not only did I have a great trip to the Evergreen Nature Center just 15 minutes south of our DeepScience office, but I was able to get in a short hike along the side of Evergreen Lake, plus the generous assistance from Evergreen Audubon that we needed for our project.
Now for a small bit of technical detail - models are only so accurate and our goal was to achieve 95% accuracy. With 5 bird species we managed to get 95% accuracy, but with 20 species we were once again below where we wanted to be. It was expected since with 20 species there are smaller differences between the species than with 5. We will have to improve our model further. It will be a challenging problem to solve but are committed to finding a solution!
Economic Opportunities Around What We Do
I opened up our office in Evergreen because I wanted to see economic development here that is in line with what people living here want in their community - development that is quiet, sustainable, and high-paying. There are few things that can fit the bill quite as well as information technology, particularly the fields around big data and artificial intelligence.
When a graduate from Carnegie Mellon's computer vision program has a starting salary of $200k/year, I see opportunity. I see opportunity when people in our advanced AI study group are earning upwards of $150k a year simply by virtue of the field they happen to work in. I personally managed to double my salary twice in 5 years by learning big data and artificial intelligence, then proving it on increasingly more complex projects. The opportunity for economic development in Evergreen from these two fields alone is unparalleled.
Big data and AI - you guessed it - have far more applications than bird identification. They are used for self-driving cars, object detection from satellite images, automated diagnoses of diseases from medical images, and improvements to cybersecurity to name a few. It is a whole new world of opportunity and one I'd like to introduce my community to.
Getting in Touch
DeepScience mainly does contract engagements on problems of scientific and technological significance, using big data and AI as tools and sometimes as solutions. We do R&D, software development, and technology consulting and would love to sit down with you in our office in Bergen Park to learn more about your needs. We're at 32135 Castle Ct, Suite 208 in Evergreen. Call us at 303-800-5707 or use our website at https://deepscience.ltd to book an appointment today.
May Blog Queen
Free Summer Recreation Around Evergreen — 5 Top Parks, Playgrounds
By Julie Dikken
Owner, Macaroni Kid