Well after almost a month, I am back from field work (and some holiday!) in Australia. I was there conducting pilot studies for a new research project I am developing. The aim of this trip was to test two methods I could use to track fish underwater using inexpensive cameras. Since manually analysing video takes sooooo long, I had to set up my cameras so that a computer could track an individual fish. I also wanted to use the same video to build a 3D model of the terrain the fish was swimming in so I could get an idea of the navigational routes the fish were taking. To get 3D information from video, we filmed using two cameras placed side by side. This creates a stereo system that acts the same way as your two eyes to give you depth information. If you overlap two images from the cameras, you can tell how far away an object is by how closely they overlap. Objects that are far away will appear in almost the same position, while ones that are closer will hardly overlap at all.
A series of cameras on tripods were set up around the territory of a fish. In theory, as the fish travels within the area that the tripods cover, we should be able to follow them. The pro of this method are that we can put the cameras in and leave them for a few hours and we don’t scare the fish by being there. The cons, as we discovered, were that it was a bit tricky to set up at first, it requires a lot of cameras to cover a large area (I had 16 cameras) and the area you could cover was limited by the visibility conditions which could change while you were filming.
We ended up setting the cameras up on the beach first so we could find the best configuration to cover the most area while having overlap with the other cameras. Once we got the hang of setting it up, this method captured some great fish activity.
Here is what the footage looks like and a few of the visitors we captured:
Using only two stereo cameras, a snorkeler followed an individual fish for about an hour. The potential drawback is that the observer can scare the fish and change its natural behaviour but I found that it only took about 5 minutes or less before the fish got used to being followed. The benefit was that we can see the fish regardless of the visibility conditions and we are less likely to lose sight of the fish if they go behind a rock. I also like this method a lot because you can cover more distance. The analysis is likely to be the tricky part but we have yet to start.
Here is what the footage looks like:
Here are a few photos of our time on the island.