TheSpaghettiDetective / obico-server

Obico is a community-built, open-source smart 3D printing platform used by makers, enthusiasts, and tinkerers around the world.
https://obico.io
GNU Affero General Public License v3.0
1.46k stars 295 forks source link

Time-lapses: timeline chart with clickable detection points to focus playback #671

Open peterlanoie opened 2 years ago

peterlanoie commented 2 years ago

Context: I have had several prints that came up fishy with false positives on infill. It would be helpful to see specifically where (when in time) those get flagged instead of blindly scrubbing through the time lapse looking for the meter spikes and highlights/indicators.

Problem: When viewing a time-lapse of a detected failure (particularly those that don't pause or are successful but smell a bit fishy), there's no easy way to scrub through the video to see where a failure is detected. The focused feedback captures are useful and can give you some hint to where in the print it has detected something, however, those don't highlight what it's detecting. The full screen view of the video does also help (widens the playback/scrub line), but for a long print, you still may not be able to find where something was flagged.

WheresTheFailure

Solution: Good: A list of timestamp points in the video that can be clicked to "auto-scrub" to that point in the video. Better: A timeline or some kind of graph of the points in time of failure detection with click points.

Here's a very badly drawn mockup. Imagine more white dots along the red timeline chart to focus on points of detection.

image

olivierberthomme commented 1 year ago

Much interested in this feature. The graph could also give a feedback that the AI process works (I always wonder if the ml container has enough power).