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
Is the added 'cleanness' of having VPR.ref and VPR.query as attributes of the VPR class worthwhile when we already should have all the data in the included VPRImageProcessor object, e.g. VPR.vip.IMG_FEATS, VPR.vip.ODOM, etc?
If not, should the VPR.set_query_matches method for calculating the true query matches (e.g. based on odometry distance or file index) be moved into VPRImageProcessor?
Do we want to have a unified Integrity.train() method, or do we anticipate training setup being different enough for each integrity subclass that we would need to rewrite it with different inputs every time?
Could we just add distance and match calculation into VPRImageProcessor and not have a separate VPR class that subclasses are based on?
Create a high-level UML diagram to describe the interfaces of the VPR class.
The text was updated successfully, but these errors were encountered: