Running BB

[mixmixmix]

@aakanksharathore, continuing discussion from #17

You are right, output videos are indeed fast in the beginning and then normal speed for the specified duration. The problem is that till now we were not writing actual frame number in the outpyt file so we cant see this pattern in text files. Will send you output files in the morning they are on another system in the lab.

That is great! If you send me the output files and videos.yml used to create them then I think I can work out correct output :}.
The step_size was set to 15 in all those runs? Right?

[mixmixmix]

I have merged all the changes with your branch bb_tracking and pushed to github. There is a change in how we pass arguments, but that should be clear from Readme.md and test.sh.

[mixmixmix]

I'm now running videos from you with step size 5 (equivalent of 6fps) (to see if they take reasonable time)

[aakanksharathore]

For full duration or specified time? I guess you can do testing for the specified 5 mins for step size 1 i.e. all frames for those 5 mins, then we will get an idea of how much time it takes at euclid compared to our servers. *Namastey!* Akanksha Rathore PhD scholar CES, Indian Institute of Science Bangalore 560012 website: https://sites.google.com/view/akanksha-rathore twitter: akanksha_ak5 <https://twitter.com/akanksha_ak5>

…

------------------------------------------------------------------------------------------------------------- सर्वे भवन्तु सुखिनः सर्वे सन्तु निरामयाः । सर्वे भद्राणि पश्यन्तु मा कश्चिद्दुःखभाग्भवेत् । ॐ शान्तिः शान्तिः शान्तिः ॥

On Wed, Jun 26, 2019 at 3:07 AM M. Kunde ***@***.***> wrote: I'm now running videos from you with step size 5 (equivalent of 6fps) (to see if they take reasonable time) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#19?email_source=notifications&email_token=ADJAAIXTYE5X6CDC3ETLR33P4KF3NA5CNFSM4H3LWLL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYRVN5A#issuecomment-505632500>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADJAAIUSW6GOVBPR66JA5NTP4KF3NANCNFSM4H3LWLLQ> .

[mixmixmix]

ok!
I did 20 and 21st march so far (at step size = 5) and it took about 2hrs per video. Now running with step size = 1 as you recommended. I will be able to copy videos and make sure that all work possibly tomorrow when I'm at the uni

[aakanksharathore]

Thanks so much! *Namastey!* Akanksha Rathore PhD scholar CES, Indian Institute of Science Bangalore 560012 website: https://sites.google.com/view/akanksha-rathore twitter: akanksha_ak5 <https://twitter.com/akanksha_ak5>

…

------------------------------------------------------------------------------------------------------------- सर्वे भवन्तु सुखिनः सर्वे सन्तु निरामयाः । सर्वे भद्राणि पश्यन्तु मा कश्चिद्दुःखभाग्भवेत् । ॐ शान्तिः शान्तिः शान्तिः ॥

On Thu, Jun 27, 2019 at 2:10 AM M. Kunde ***@***.***> wrote: ok! I did 20 and 21st march so far (at step size = 5) and it took about 2hrs per video. Now running with step size = 1 as you recommended. I will be able to copy videos and make sure that all work possibly tomorrow when I'm at the uni — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#19?email_source=notifications&email_token=ADJAAIV2OWK4GPC53X2VNELP4PH3RA5CNFSM4H3LWLL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYUYAQI#issuecomment-506036289>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADJAAIR3KHSLF5HYSEOXWMLP4PH3RANCNFSM4H3LWLLQ> .

[mixmixmix]

So, I have first three almost done after 12 hours.
So far 24k frames, so the average processing is about 30 frames per minute.
How fast is it on your server?
Once I'm at the uni I can send you videos for Skip5 and Skip1 for comparison, and start the rest o the videos at Skip5 on Euclid - so if subsampled version is not good in some cases we can run it later but you'd have something to work with :)

[aakanksharathore]

Hi Miks, 30 Frames per minute is great. For me, it takes 24 hrs and sometimes even more to process 5mins of videos at Skip1. Can you please run remaining videos at Skip1? We are anyway tracking only 5 mins of each video, we had compared skip5 and skip1 when you guys were here and decided to go with skip1. *Namastey!* Akanksha Rathore PhD scholar CES, Indian Institute of Science Bangalore 560012 website: https://sites.google.com/view/akanksha-rathore twitter: akanksha_ak5 <https://twitter.com/akanksha_ak5>

…

------------------------------------------------------------------------------------------------------------- सर्वे भवन्तु सुखिनः सर्वे सन्तु निरामयाः । सर्वे भद्राणि पश्यन्तु मा कश्चिद्दुःखभाग्भवेत् । ॐ शान्तिः शान्तिः शान्तिः ॥

On Thu, Jun 27, 2019 at 1:49 PM M. Kunde ***@***.***> wrote: So, I have first three almost done after 12 hours. So far 24k frames, so the average processing is about 30 frames per minute. How fast is it on your server? Once I'm at the uni I can send you videos for Skip5 and Skip1 for comparison, and start the rest o the videos at Skip5 on Euclid - so if subsampled version is not good in some cases we can run it later but you'd have something to work with :) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#19?email_source=notifications&email_token=ADJAAIVHG3C7TTPOAMPR5BLP4RZYNA5CNFSM4H3LWLL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYWK2FY#issuecomment-506244375>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADJAAIUEIE3GYMOXNOBSZ53P4RZYNANCNFSM4H3LWLLQ> .

[mixmixmix]

I will need to run tracker on my data in a few days, so I can run and see how far I get with your data until then.

What is your server's setup?

[aakanksharathore]

No problem, Miks. I have also asked for access to the divisional server, I don't know yet how fast it would be. Here is the configuration of nodes on which I am running it currently- [image: image.png] *Namastey!* Akanksha Rathore PhD scholar CES, Indian Institute of Science Bangalore 560012 website: https://sites.google.com/view/akanksha-rathore twitter: akanksha_ak5 <https://twitter.com/akanksha_ak5>

…

On Thu, Jun 27, 2019 at 4:35 PM M. Kunde ***@***.***> wrote: I will need to run tracker on my data in a few days, so I can run and see how far I get with your data until then. What is your server's setup? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#19?email_source=notifications&email_token=ADJAAISU7BDHFBQSRVRJ3DLP4SNHTA5CNFSM4H3LWLL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYWYX5I#issuecomment-506301429>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADJAAITRW7VPGYV3OTH74PTP4SNHTANCNFSM4H3LWLLQ> .

[mixmixmix]

The image didn't work, could you post it directly through the github?

[mixmixmix]

I'm half-way through with videos. I will try to upload them somehow for you around the weekend (all skip5 and those skip1 which have finished)

[aakanksharathore]

Thanks, Miks. I am also running some videos here but of course it is taking around 2 days for 5 min video at skip1. Are you also getting some ID fluctuations in the stationary part of videos? *Namastey!* Akanksha Rathore PhD scholar CES, Indian Institute of Science Bangalore 560012 website: https://sites.google.com/view/akanksha-rathore twitter: akanksha_ak5 <https://twitter.com/akanksha_ak5>

…

On Thu, Jul 4, 2019 at 7:49 PM M. Kunde ***@***.***> wrote: I'm half-way through with videos. I will try to upload them somehow for you around the weekend (all skip5 and those skip1 which have finished) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#19?email_source=notifications&email_token=ADJAAIWWTCNTSHLZLTYK4GDP5YBHVA5CNFSM4H3LWLL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZHRIFQ#issuecomment-508498966>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADJAAIQKVQHXPUEBPAOOHLTP5YBHVANCNFSM4H3LWLLQ> .

[mixmixmix]

https://youtu.be/sMf9-c7rZ8g - like that? This is what the issue 22 was about (#22)
I don't think running at skip1 is a necessary. I will upload all the skip5 I did last week, and what skip1 I have so far (there are 6 videos left).
Can you provide the spec of your server again? I remember someone saying that you have nvidia 1080 on one of the computers?

[aakanksharathore]

No, in my output videos there aren't that many fluctuations. In our discussion with Vishu and Colin, we decided to go with skip1 since when we analyze the data, skip1 is giving the results with less noise. *Namastey!* Akanksha Rathore PhD scholar CES, Indian Institute of Science Bangalore 560012 website: https://sites.google.com/view/akanksha-rathore twitter: akanksha_ak5 <https://twitter.com/akanksha_ak5>

…

On Fri, Jul 5, 2019 at 8:45 PM M. Kunde ***@***.***> wrote: https://youtu.be/sMf9-c7rZ8g - like that? This is what the issue 22 was about (#22 <#22>) I don't think running at skip1 is a necessary. I will upload all the skip5 I did last week, and what skip1 I have so far (there are 6 videos left). Can you provide the spec of your server again? I remember someone saying that you have nvidia 1080 on one of the computers? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#19?email_source=notifications&email_token=ADJAAIVRNTYQM5L46SV327DP55QS3A5CNFSM4H3LWLL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZJYPYQ#issuecomment-508790754>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADJAAISCH6GX5HHWQAMCJIDP55QS3ANCNFSM4H3LWLLQ> .