I am a wildlife biologist working on a project to survey birds using cameras.
We took thousands of high-quality aerial images of flocks of (large) birds on water. In most images, the birds are VERY easy to detect by eye and, I THINK, would be very easy to isolate and detect by computer program. I mostly use R, and R is not great for this type of project and I don't have time to learn other software. A past employee with moderate coding experience largely already did this in matlab, but retired before he could fix some glitches. However, I'd like a program that I can run without the use of an expensive program like matlab.
For the current year, we have about 12,000 photos. And we will re-do this survey every year. So, Ultimately, I'd like a final product that is fairly easy to run by almost anyone with some coding/computing skills. There are several species in the images and if remotely possible, it would be nice if the program could try to sort them. For instance, if it could give you a likelihood that item x is a bird and the likelihood that it is species x vs. y.
I included a photo of a flock of birds on water from the project.
There are a few other issues -- but I don't think they would be very difficult. For example, some photos are over land, not water, and can be thrown out. There are other "items" in the pictures like rocks, garbage, other wildlife (e.g. seals), and the program would need to not count these.
Let me know if you think this is possible!
21 freelancere byder i gennemsnit $2437 på dette job
Hi I am very interested in your project. I have strong experiences in Image Processing. I am looking forward to discussing your project. Best Regards.
I have worked on object detection using CNN in Python and Tensorflow. The idea was to count and detect vehicles in images. Relevant Skills and Experience Python, Tensorflow, CNN (Deep Learning)