It would be useful to pair Tomnod with http://www.terrapattern.com/. For example, rather than scanning many empty areas of ocean looking for a lost boat, find one example pattern and quickly come up with geographically market best possibilities.
One good thing about scanning empty ocean areas is finding all the “faces” and other shapes our brains find as they try to “see” something recognizable in the waves and swirls - fun finds. Although it may be boring to look at so many empty map tiles, once in a while our brains will trick us into seeing something unusual (that’s not really there). Sometimes it helps break up the monotony.
I did check out TerraPattern and was impressed with the boat wakes and cul-de-sacs. TerraPattern would have been nice to try out on the swimming pools in Adelaide campaign!
One problem, stated in the FAQ, is “The Terrapattern project uses tile-based search, not pixel-based search. In other words, the tool finds places that are similar to the map tile you clicked in—not the exact object you clicked on. As a result, the system may be serving up tiles that match your selection in ways that are different than the ones you’re expecting.”
So, it’s accuracy would not be as high as visually searching. Terrapattern may be more like what Tn is doing to teach the computer to recognize shapes, which is why we humans assist with marking the computer’s polys for accuracy. Just guessing, I suppose it could take billions of the humans’ ratings before computers learn to detect shapes and objects versus shadows. However, Tn’s computer algorithms are often fairly sharp, such as when it spotted just corners of cars parked under trees in a recent campaign! Of course, it still misread a tree’s leaf edges as “corners”-- no car, just a tree. Or, when it picks out painted markings on streets, and marks them as vehicles… I always shake my head and say, “No, Server Johnny, not a car!” LOL (Just how do you teach a computer these nuances?!)
Right. I am thinking that Terrapattern would be used for prioritization of tiles, rather than as a replacement technology. So, find where the computer thinks boats or cars or pools are most likely (especially for rescue efforts that are time critical). Humans would still do the checking on those and then on the other tiles.
I love where you’re going with this @Peter8 -& Terrapattern is awesome! You may have noticed some of our campaigns asking the crowd to help validate algorithm results & help build training data. Most of these tasks are to support an internal movement to incorporate machine learning in order to improve the accuracy and scalability of crowdsourcing (we call this crowd + machine).
We’ve been testing and building this initiative with relatively easy features such as swimming pools & solar panels. There are others in the works as well, such as a boat detection initiative, which could potentially help during marine-based SARs.
It’s not an easy thing to accomplish, which is why it’s important to learn from others in the community, such as Terrapattern.
The goal, for Tomnod, is that instead of asking Nodders to search 100,000 km2 of imagery, we would instead ask them to simply validate the results from a machine learning image mining algorithm -one that has been trained with Nodder input
Just tell him you’re going to spank his port. (USB port that is!)