Mosquitoes are very small, fast moving and difficult to see. You will need a very good camera to pick one up! Tracking one will be a very difficult project.
A problem with fish eye cameras will be how to compensate for the fish eye distortion. It sounds like you will need to understand the real world position of the detected object. If you detect an object at pixel (x, y), where is that object in real space?
A more expensive fish eye designed for computer vision type applications may have a mapping for this, or you can try to find it yourself. A normal lens camera can be approximated as a pinhole camera which makes this much more simple.
In terms of resolution, you can work our your requirements from the desired sensitivity of your system. Have a look at this:
https://www.scantips.com/lights/subjectdistance.html
Be careful though, many cheaper cameras have a high pixel resolution but have a lot of pixel crosstalk and so the effective resolution is much lower. Also how many pixels do you need to detect the object? Depends on the method you use.
Also the intel camera includes a depth sensor. Without a depth sensor you will only know the direction from the camera, and not the distance. Also depth information can help detection, depending on your method. Be careful though, the resolution of the depth map will often be less than the camera, and mosquitoes are very small, I doubt it will pick one up.
I've not used this intel camera but it looks like there is a lot of support for using it for these types of applications which is also a bonus.
In my experience the quality of the camera is the biggest factor that affects performance, more than algorithm choice.
No real answers there but just some thoughts.
