I ran the example doors given and it missed 9 swinging doors, some that were in double swing pairs, and a few that were just out on their own not clustered. Not bad overall though

▲

wcisco17 4 hours ago | parent [-]

Yep we're constantly improving we're currently above 0.87 for doors

we're thinking of adding a params for the ROC curve so that you can decide your own optimal thresholds depend on when false positive true positive rate is acceptable

	▲	petee 3 hours ago \| parent [-]
		Oh nice! I spend a good amount of time eyeballing drawings for overlooked details, so even finding most is a handy tool to me as my brain can skip the marked areas