Hi,
It has been stated that the segments in the bacground set are "signals in the same geolocation, time and direction that did not receive the target", and that "The background's segments have segment_id with 1,000,000 added to the original segment_id in the training set or the auxiliary experiment set".
But it seems I get segment/track id's that appear in the background set twice, with different geolocations id/type and date.
The first seems to match a track from the training (or experiment) set but I don't understand where the second one comes from. Could you help?
As an example, background_set.iloc[0]:
segment_id 1000421
track_id 349
geolocation_type A
geolocation_id 2
sensor_id 2
snr_type HighSNR
date_index 5
target_type empty
And, background_set.iloc[30889]:
segment_id 1000421
track_id 349
geolocation_type C
geolocation_id 5
sensor_id 5
snr_type LowSNR
date_index 4
target_type empty
Hi,
Yes, you are right. It seems that some of the background segment IDs are incorrect. We will post a patch to the background segments dataset as soon as possible.
Thank you for your notice,
Or (MAFAT challenge team)