Comparative analysis for detector using dataset 2

[sfmig]

We can compare performance between a few different scenarios:

• [ ] off-the-shelf vs additional channel
• [ ] if additional channels looks promising, then check different variations of additional channels
• [ ] is adding dataset 1 to the training set useful?

If we get here and dataset 2 is not labelled, then start off with dataset 1

Blocked by

• 69

[samcunliffe]

Potentially not enough days labelled yet for an interesting study. Moving on pause for now.