feat: add dgp2wicker conversion #132
Conversation
chrisochoatri
force-pushed
the
dgp2wicker
branch
4 times, most recently
from
51ab0dd to
c0a2a0d
Compare
chrisochoatri
force-pushed
the
dgp2wicker
branch
2 times, most recently
from
d81e0d7 to
419b291
Compare
chrisochoatri
force-pushed
the
dgp2wicker
branch
2 times, most recently
from
e187bd6 to
23a23ca
Compare
There was a problem hiding this comment.
Reviewed 11 of 16 files at r1.
Reviewable status: 11 of 16 files reviewed, 4 unresolved discussions (waiting on @chrisochoatri)
dgp/contribs/dgp2wicker/Dockerfile line 1 at r1 (raw file):
FROM dgp:latest
copyright
dgp/contribs/dgp2wicker/sample_wickerconfig.json line 1 at r1 (raw file):
{
copyright
dgp/contribs/dgp2wicker/dgp2wicker/cli.py line 117 at r1 (raw file):
) print(results)
wait, just print?
dgp/contribs/dgp2wicker/dgp2wicker/dataset.py line 224 at r1 (raw file):
""" wicker_samples = self.wicker_sample_index[index] idx = wicker_samples[0]
I assume this shouldn't always get 0?
If yes, put a comment here to describe what's inside wicker_samples (or why 0)
dgp/contribs/dgp2wicker/dgp2wicker/ingest.py line 351 at r1 (raw file):
Number of partitions to shuffle all samples over. If none, defaults to num_scenes*5 is_pd: bool, default=False
btw I don't like this way but it seems we don't have better way lol
There was a problem hiding this comment.
Reviewable status: 11 of 16 files reviewed, 4 unresolved discussions (waiting on @kuanleetri)
dgp/contribs/dgp2wicker/sample_wickerconfig.json line 1 at r1 (raw file):
Previously, kuanleetri (Kuan-Hui Lee) wrote…
copyright
can't, no comments in json file. This is just intended as an example wicker config that the users can update with their own s3 path.
dgp/contribs/dgp2wicker/dgp2wicker/cli.py line 117 at r1 (raw file):
Previously, kuanleetri (Kuan-Hui Lee) wrote…
wait, just print?
This will print something like {'train' 50000, 'test':1000} etc, should I add more verbose output?
dgp/contribs/dgp2wicker/dgp2wicker/dataset.py line 224 at r1 (raw file):
Previously, kuanleetri (Kuan-Hui Lee) wrote…
I assume this shouldn't always get 0?
If yes, put a comment here to describe what's insidewicker_samples(or why 0)
good catch. This was a typo left over from testing, I removed it.
dgp/contribs/dgp2wicker/dgp2wicker/ingest.py line 351 at r1 (raw file):
Previously, kuanleetri (Kuan-Hui Lee) wrote…
btw I don't like this way but it seems we don't have better way lol
I agree, not ideal...
chrisochoatri
force-pushed
the
dgp2wicker
branch
from
23a23ca to
3916264
Compare
There was a problem hiding this comment.
Reviewable status: 10 of 16 files reviewed, 4 unresolved discussions (waiting on @kuanleetri)
dgp/contribs/dgp2wicker/Dockerfile line 1 at r1 (raw file):
Previously, kuanleetri (Kuan-Hui Lee) wrote…
copyright
Done.
chrisochoatri
force-pushed
the
dgp2wicker
branch
from
3916264 to
6223430
Compare
chrisochoatri
force-pushed
the
dgp2wicker
branch
from
6223430 to
a06ec81
Compare
There was a problem hiding this comment.
Reviewed 2 of 16 files at r1, 2 of 2 files at r2, 3 of 3 files at r3, all commit messages.
Reviewable status:complete! all files reviewed, all discussions resolved (waiting on @chrisochoatri)
This adds a semi standalone docker for saving DGP Synchronized Scene output in Wicker.
This is tested primarily manually due to the need for s3 access. Please build the docker, pass s3 credentials, and run test.py for more details.
This change is