Parrot and use case to recognize and identify objects in real time

I have a use case to identify and count cattle in fields. In the past, using non parrot drones, we have flown over the fields, collected the photographs, used photogrammetry, and then passed the resultant images through tensorflow files to identify and count the number of animals in a field. The work flow is bulky and requires a large amount of data to be moved and then processed.

I just saw that Parrot has both the groundsdk and airsdk and now can communicate via LTE with verizon - at least this set up is coming in Jan 2022.

Can the parrot ecosystem, between the groundsdk and airsdk, complete this use case without having to collect, move, and process all this data? For instance, can the airsdk grab video frames, pass the frames through a tensorflow file while on the drone, identify the objects, using location information place those objects at specific coordinates on the ground, and then only pass back the location and number of objects to the ground sdk and eventually to a cloud storage via the LTE?

I’m not looking for specifics just if the parrot ecosystem can accomplish this use case? We have been wanting to have this capability for a long time.

Thanks in advance,

This topic was automatically closed after 30 days. New replies are no longer allowed.