difference 6K Global Model vs V2.4 FP 16

[pddpauw]

Hi All,

objective of this note is to give some perspective on the differences between the 2 models.

I have 2 Pi's, with similar microphone, 10 cm away from each other.
Pi 1 is running the 6K Global Model
Pi 2 is running the V2.4 FP16 Model (species occurency frequency threshold 0,0006)

Below the detections for Pi 1 with 206 detections and 34 species detected:

Below the detections for Pi 1 with 549 detections and 54 species detected:

Pi 1 has several false positives which either don't live here (obvious of course, as it doesn't take location into account), or are definitely not clear/correct. E.g.

• Muscicapa striata: should be in Africa now (although the sound was very similar to this bird)
• Strix uralensis: was rather the barking of a dog
• Botaurus stellaris: was only some noisy sounds
• Tyto alba: was in fact the sound of an alarm
• Aegolius funereus: should have been the Strix aluco

On the other hand, the Pi1 detected a bird which I couldn't find on the Pi2:

• Bombycilla garrulus: relatively sure this was correct (was during couple of days where it was freezing). No visual confirmation though.

So, I will put the V2.4 back on!

Pieter