DDAD trainval is so big, is there a way to download some small videos and calibration params?

Question

lucasjinreal opened this issue 4 years ago · 13 comments

Answer 1 · 2021-06-05T02:10:38.000Z

We have a tiny version of DDAD in the following link (it's used in packnet-sfm to run overfit tests).

s3://tri-ml-public/github/packnet-sfm/datasets/DDAD_tiny.tar

Answer 2 · 2021-06-05T15:45:31.000Z

@VitorGuizilini tiny only contains less than 10 images.

Answer 3 · 2021-06-08T14:38:57.000Z

@jinfagang I can generate a small version of DDAD train with about 10 scenes -- would that help?

Answer 4 · 2021-07-26T09:39:58.000Z

@RaresAmbrus Yes. that definitely helpfull! thank u!

Answer 5 · 2022-09-14T19:40:04.000Z

@RaresAmbrus Could you please tell me where can I download the DDAD train dataset with 10 scenes which you had generated?

Answer 6 · 2022-09-19T05:52:09.000Z

Hi, we have a "tiny" version of DDAD as part of the packnet_sfm repo, does that work: https://github.com/TRI-ML/packnet-sfm#tiny-ddadkitti ?

Answer 7 · 2022-09-19T11:47:26.000Z

@RaresAmbrus tiny version is very small, it has only a few images.

Answer 8 · 2022-09-27T01:19:54.000Z

I created a smaller version here -- this one has 10 scenes and is about 13GB. Can you please check if you can access it? Thanks!

Answer 9 · 2022-09-28T03:02:49.000Z

Yes, I could access it. Thank you so much.

Answer 10 · 2022-09-28T03:16:40.000Z

One more thing, is there a way to do inference on single image or folder using model trained with Vidar repo?

Answer 11 · 2022-10-06T17:36:49.000Z

can anyone please reply to my query?
Thanks in advance!

Answer 12 · 2022-10-06T17:51:25.000Z

Hi @akashchavan15 - sorry I missed your message. Can you please post that on the vidar repo? Thanks!

Answer 13 · 2022-10-06T18:00:43.000Z

@RaresAmbrus thank you for your reply. I have just posted it on Vidar repo. Here is the link
TRI-ML/vidar#26

Please do reply.