In the demo notebook in the 'Load the dataset' cell I'm missing a dataset about synsin HOT 16 CLOSED

I don't mean to be rude, but I don't understand why I have to labor over this for 6 hours. Where does the dataset go? Why do you make users solve these puzzles instead of just making it work out of the box?

Can I have a clue? I've tried the dataset in the data folder and the root folder of the repo.
I've done what you said. I figured how to comment out the .txt file and I got the dataset into the repo. Can I have another hint please?

from synsin.

Let's get things clear here. Ok you are having difficulties with the repo. Life is hard. You will figure it out. But if you try again to fish for comments with some article or ask folks where they work, which is none of your goddamn business, I will have you blocked from all facebookresearch github repos. And yes you are being rude.

from synsin.

The video_loc.txt we created in order to deal with frames or videos that may be removed from YouTube. It simply contains the successfully processed frames in the text file. This is described here: https://github.com/facebookresearch/synsin/blob/master/REALESTATE.md. You can of course modify how this works to not require the video_loc.txt

from synsin.

You can of course modify how this works to not require the video_loc.txt

Do I need this dataset to run the Interactive Demo notebook?
Where do I download it to?
What do I modify to not require video_loc.txt?

from synsin.

To run that notebook, yes you need the dataset. The instructions are in the linked file above.

To run the simpler demo: https://github.com/facebookresearch/synsin/blob/master/demos/Simple%20Demo.ipynb. You do not need any dataset.

from synsin.

There are no instructions on where the notebook wants the dataset to be in the link.
There is nothing about where to " modify how this works to not require the video_loc.txt".

The dataset wouldn't download via the command line, so I worked around that.
I guess I'm putting it in the root directory of the repo and commenting out

#         self.imageset = np.loadtxt(
#             opts.train_data_path + "/frames/%s/video_loc.txt" % "train",
#             dtype=np.str,
#         )

in the file
http://localhost:8882/edit/data/realestate10k.py

Most notebooks just download the dataset if it's not stored locally.

Is my stab in the dark near what has to happen to run the notebook?

Do you work at Facebook? Have you seen this? What do you think about that?

from synsin.

Unfortunately, we can't just make the dataset available to be downloaded due to IP reasons; hence we explain how to download the dataset in the readme file . This is why it has to be downloaded by the user following these instructions. If you don't want to have to download the dataset, you can use the 'Simple Demo' notebook I linked above. This requires no dataset.

I am a PhD student. I no longer work at Facebook.

from synsin.

I set things up like I described in my last comment.
I'm getting
AttributeError: 'RealEstate10K' object has no attribute 'imageset'

In the same 'Load the dataset' cell in the 'Interactive Demo' notebook.

from synsin.

I'm sorry. I didn't mean to cross boundaries.

If I got more than "it's in the link" (it wasn't in the link) or "you'll figure it out", people with the same problem with this new repo in the future would be able to see my Q and your A, that is, if it answered the question.

As far as my question about Facebook, it's clearly some kind of military technology deployed for counter-insugancy purposes against the now locked down public and Mark Zuckerberg is an actor.
I'd appreciate it if you did NOT block me. I guess you can't talk about that other stuff because it would be 'on the record' here.

from synsin.

@oawiles has given you instructions on how to download the dataset. @oawiles also explained why we can't provide the dataset ourselves (tl;dr legal issues). So yes, you need to try a little bit to download the dataset in order to use the demo. I understand that some might find it harder than others to set it up, and we are happy to answer questions.

The rest of your questions and mentions are completely irrelevant. You keep insisting on mentioning things unrelated to this repo, and your commentary on "on the record" statements is really pushing it for me. One more strike, and you're blocked.

from synsin.

I understand that some might find it harder than others to set it up, and we are happy to answer questions.

Okay, sorry. I got that far.
I have the dataset and I commented out the video_loc.txt lines.
I have the dataset in a folder called RealEstate10K with train and test dirs inside of it.
I put the folder in the main directory. I copied the folder into the data directory.
I get AttributeError: 'RealEstate10K' object has no attribute 'imageset'

Uhhhhh. Is the object the RealEstate10K class?
What is supposed to have the attribute imageset? I know this might sound sarcastic, but I am asking "What is supposed to have the attribute imageset?"

from synsin.

I know I'm down to the issue of where the folder of the RealEstate10K goes.
I see it's looking in some place opt.train_data_path =("/checkpoint/ow045820/data/realestate10K/RealEstate10K/")
in /options/options.py

from synsin.

You can of course modify how this works to not require the video_loc.txt

Can we go back to focussing on this step?

from synsin.

https://github.com/facebookresearch/synsin/blob/master/REALESTATE.md

"The frames need to be extracted based on the text files; we extract them to: ${REAL_ESTATE_10K}/frames"

With what? How?

from synsin.

Are you really happy to answer questions?
Do I use youtube-dl to download each video and ffmpeg to convert each frame to png?
I have never seen anything like this.

from synsin.

THanks

from synsin.