Experiments

[sbdzdz]

Data:

• improve scale and position normalisation
• add orientation alignment
• add a random map-style dataset

Models:

• add a parameter conversion method
• improve logging callbacks
• refactor training scripts
• add a joint training scenario

Visualisation:

• add an orientation normalisation animation
• add joint training loss plot
• improve continual training loss plot

Other:

• refactor
• rename some datasets
• add test scripts

Joint training: