Joseph Nelson
speaker
239 appearances
1 recordings
1 series
first heard Dec 2025
last heard 18 Dec
Joseph Nelson’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 1 in all, peaking in Dec 2025 with 1.
Appearances
So it just does uh detection of of the task.
So you can envision that the same way the SAM3 paper introduces the idea of using SAM3 in tandem with MLLMs.
I would expect that to be the case.
Pretty soon and maybe the Google team
taking some notes to improve.
Gemini and other series of models based on what SAM three demonstrates here.
So in other words, not only is it faster, but it seems to be more comprehensive for concept segmentation.
The other interesting thing is the Florence 2 results.
And you know, Florence 2 is a little bit older of a model now, so maybe it's not fair to put up head to head with the state of the art, but it is useful as a way to just see how far we've come.
Because Florence 2, by comparison, labels the entire region as uh a single class without seeing individual detection of numbers and indicator lights and needle.
And not only that, but it actually runs at about three times
The speed is is SAM three.
So CM3 again is faster doing a task that the other models are not doing in segmentation and more accurate, both in recall and precision, of the things that it's intended to find, which I think really showcases the capabilities of of the model.
encouraging.
Even a task that wasn't expressly
prioritized, it still does a great job on
an easier one.
Glass mug, Sam three, Gemini three, Florence two.
SIM three loaded first and has
Really impressively, it sees even this glass mug in the corner, which I think is something SAM3 does a great job of is occlusion and partial objects.
Showing 121–140 of 239 · page 7 of 12
← Previous
Next →