A protocol for Team Video is in the early stages, and there are many aspects to consider. While it started slow, we have found a relatively quick and efficient system for placing cameras, taking down and storing cameras, and uploading videos. Unfortunately, many additional challenges await.
Watching the 2+ terabytes of footage will be a long and arduous task, and it is therefore key to plan well in these early stages. I believe the most important questions to ask at this phase are:
What data are we looking to collect?
With so much footage we have captured a lot of different things that we could potentially measure. We can see thrips on nearly every head we’ve recorded. We also have seen many ants, though typically of only two main species. Previous hypotheses about the roles of these ants have been posted, and the videos of the flowering heads would be a great resource for anyone wanting to find out more about them. As of now (though I’m not entirely sure as I haven’t thoroughly consulted with the rest of Team Video) I believe we are only going to record information about the insects that visit the heads and not their permanent residents.
What is the most effective/efficient way to collect these data?
This project cannot be done quickly or easily. Every day that we record in the Common Garden we get about 7-9 hours of footage per camera. With ten cameras rolling that is 70-90 hours of footage to watch per day. If we record for 5 days in a week (not uncommon) we then have 350-450 hours. That’s a freakin’ lot of video! Efficiency is key, but as a wise man once said “We’re looking to increase efficiency without losing accuracy”.
It is at these questions that I hit a wall, and would appreciate the input of my fellow bloggers/fans of the blog. As of now we have been using a video player that has the capacity to fast forward at speeds up to 32x, where should the line be drawn? At some point well below the 32x speeds, we might start missing things. Some bees only remain on the flower for a second or two before leaving, and the watcher of these videos may not catch these visits. However, there is really no way to watch these videos at normal speeds and expect to finish before the end of the decade (let alone the end of the summer). Also, a question directed at fellow Echinacea team members; are there data that you would like to be gathered from these movies that I haven’t mentioned? The ball has been passed into your court…
Hello! It’s an interesting problem to be sure. You could consider some sort of motion sensor which would turn on/off the cams when a bee or insect came across the field of view. But, that might not be fast enough to capture the movement and insect’s path.
Alternatively, if you know someone who can program a scan software program, any significant percentage difference in several frames could result in marking and saving that video to hard drive or dvd burner.
Given the quantity of video, this might be the best course. The programmer may have to build a filter to account for waving flowerheads and recognize percent blob for a bee’s profile. The bee’s constantly shifting trajectory, with starts and stops, would be helpful, too.
Do you know someone at the university who would be interested in developing such a program? It could have broad applications actually.
Well, just a thought.
Good luck!
Can you randomly sample the video footage? I wonder how much viewing of video would essentially give you what you need to know. 5% of all video? 10%? 20%?