Incorporate last videos from Y

[brialorelle]

• Download new videos added in the last few months
• Run OP on these videos, save json outputs to both Sherlock and langcog
• Get corresponding meta-data (as avaliable)

[ketan0]

New Y videos are currently downloading to sherlock - Databrary appears to be quite a bit slower than last time for some reason (>10 mins per vid,) so it may take a while to download the larger dataset...

[ketan0]

Y finished downloading and OP is running on Y videos, with a script at the end of each job to condense the outputs for each video into a single json dataframe that can be loaded in pandas/R.

[kachergis]

awesome! can you let us know when they've finished running?

[ketan0]

Hey! With a couple small exceptions of videos that timed out/video got deleted after the job was scheduled, which I will re-run, the videos are mostly done running.

[ketan0]

Y videos are finished running! A and S videos currently running

[brialorelle]

Great, thanks for the updates Ketan! I just wrote a pipeline to redo main analyses directly form the bounding box outputs so we have some built in sanity checks. Doing things the video level make for a considerably easier time...)

…

On Tue, Jun 2, 2020 at 3:24 PM Ketan Agrawal ***@***.***> wrote: Y videos are finished running! A and S videos currently running — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABN2D4BMCUMRNJYTVFOZVJDRUV3YJANCNFSM4NBY33AQ> .

-- Bria Long, Ph.D. Postdoctoral Fellow, Stanford University www.brialong.com | 650-644-7235

[ketan0]

AFAIK, there doesn't seem to be any Y metadata available anywhere. I tried emailing Michelle Mei to confirm, but it seems the email has bounced (is she still at Skidmore?)