You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The main question is, is it required to cluster on HiRISE image level, which would solve any potential issues with overlap regions getting un-naturally higher probability to survive because they have been seen twice (or more for tile corners).
To address, we need to do both clustering ways and check if there's any clusters surviving in the P4 coordinate clustering that did not survive in the HiRISE coordinate clustering.
As a first step, a simple count comparison of defined clusters in both cases should suffice to identify any difference.
For 'ESP_011544_0985' I count 764 end objects in HiRISE coordinate space (i.e. taking care of overlap automatically) vs 809 in Planet4 tiles scope, so a rough 5% increase. Testing another image now.
Check 2 tiles next to each other and how the overlapping region is being dealt with.
The text was updated successfully, but these errors were encountered: