Hello all, and thanks for the information available.

We currently use a Phantom 3 Pro for environmental surveys.

One of our applications is counting birds on colony islands using Agisoft to stitch together grid images of colony islands into a single orthophoto, then counting the birds from the ortho.

Since some birds move between one pass and another adjacent image, they sometimes tend to be averaged out in Agisoft into a soft blur on the ortho that is difficult to identify.

We have worked on software that takes each raw photo from the P3P, rotates it according to the Yaw value at that instant in the log file, then allows you to view each raw image with the other birds already marked from previous images shown on the new image. When you touch on the image of a bird and specify its species, a mark record is generated with the Lat/Lon that can be transferred and viewed on other images. This would theoretically allow us to see individual birds in each image, and know whether it had already been marked with a previous image.

The only problem with this plan is that the GPS values associated with P3P images are just not good enough to be able to align the locations with enough accuracy to prevent double counting. P3P Lat/Lon values tend to be up to 2 meters off in some instances, and in dense colonies, that's not good enough.

With the demise of the Phantom 5, we are thinking about getting an RTK.

Higher accuracy on the GPS values might give us the ability to do away with the blurred images in an Ortho by using the rotated raw images, but the entire RTK system is quite an investment.

Our application is on islands where there is basically no topography at all – just bushes and a very occasional 2 foot sand dune.

If we can give up altitude calculations, can we use a Phantom 4 RTK without having to purchase an external base station? That seems to cut the cost in half.

All we care about is a planar Lat/Lon value with increased accuracy over a P3P, with a Yaw value at least as accurate. We also don't care that much about the absolute location of each bird on the face of the earth, only the relative location in relation to missions flow that day on the same island.

Thanks in advance for any information or suggestions

Your plan sounds fine to me. Nice solution for finding the same bird in multiple images. I have been using and testing the P4RTK for a few months and have been getting planemetric accuracy of 2-4 cm with the orthomosaics I create using Agisoft. I do this without ground control. I do use the mobile base station because I am interested in absolute accuracy on planet Earth. Yes, you could get very good relative accuracy without using the base station.

jgillan Posted at 5-20 08:41
Your plan sounds fine to me. Nice solution for finding the same bird in multiple images. I have been using and testing the P4RTK for a few months and have been getting planemetric accuracy of 2-4 cm with the orthomosaics I create using Agisoft. I do this without ground control. I do use the mobile base station because I am interested in absolute accuracy on planet Earth. Yes, you could get very good relative accuracy without using the base station.

Thanks very much

We currently use a 3rd party Android app named DJI Ultimate Flight for mission planning and autonomous flight control.

Can you either use a regular Android tablet for control, or at least run 3rd party Android apps on the RTK controller?

The controller comes with an android device attached. Currently, I am only able to use the DJI App installed on the device for mission planning. I believe the SDK has been opened up for 3rd party development but I have not seen and don't know how I would use 3rd party apps with P4RTK.

Agreed with everything jgillan said.
Also just to clarify that the "raw" images you'll be able to capture with the P4RTK will be jpeg format, not a true RAW format. Pretty sure you knew that, but just in case....