search

developmentseed / slickformer

Segmenting oil slicks with a panoptic segmentation transformer architecture
4 stars 1 forks source link

1 or more sentinel scene IDs returns access denied errors when computing reprojection parameters to associate labels with images #2

Closed rbavery closed 1 year ago

rbavery commented 1 year ago

In cli.py, we make a request to s3 to get reprojection parameters from sentinel-1 COGs. It looks like one or more scene ID's returns an access denied error.

when debugging, many scenes are correctly processed until S1A_IW_GRDH_1SDV_20200805T144239_20200805T144304_033773_03EA39_1ED1

(slickformer) root@bd2cc15c0868:/slickformer#  cd /slickformer ; /usr/bin/env /root/mambaforge/envs/slickformer/bin/python /root/.vscode-server/extensions/ms-python.python-2023.4.1/pythonFiles/lib/python/debugpy/adapter/../../debugpy/launcher 39689 -- /slickformer/ceruleanml/cli.py make-coco-dataset-no-tiles data/partition_lists/test_edited.txt data/test_coco 
Dask client dashboard link:  http://172.17.0.2:8787/status
S1A_IW_GRDH_1SDV_20210518T163942_20210518T164007_037945_047A75_22BA
S1A_IW_GRDH_1SDV_20201024T115458_20201024T115523_034938_041306_3757
S1A_IW_GRDH_1SDV_20201210T224256_20201210T224325_035630_042AF4_1998
S1A_IW_GRDH_1SDV_20201110T055833_20201110T055858_035182_041B7B_5F7D
S1A_IW_GRDH_1SDV_20210111T005559_20210111T005624_036083_043AB6_3173
S1A_IW_GRDH_1SDV_20210727T223312_20210727T223334_038969_049919_B9C1
S1B_IW_GRDH_1SDV_20210719T040729_20210719T040758_027858_0352F9_CF72
S1A_IW_GRDH_1SDV_20200802T062802_20200802T062827_033724_03E89F_6485
S1B_IW_GRDH_1SDV_20200803T173418_20200803T173443_022762_02B332_8570
S1A_IW_GRDH_1SDV_20210902T105041_20210902T105107_039501_04AAEC_9712
S1A_IW_GRDH_1SDV_20200802T174315_20200802T174340_033731_03E8D7_B2B2
S1A_IW_GRDH_1SDV_20210711T062137_20210711T062202_038726_0491D7_B424
S1A_IW_GRDH_1SDV_20200801T023915_20200801T023940_033707_03E818_D078
S1A_IW_GRDH_1SDV_20210609T023918_20210609T023943_038257_0483C1_427A
S1A_IW_GRDH_1SDV_20210306T105033_20210306T105058_036876_045642_EABE
S1A_IW_GRDH_1SDV_20210220T174437_20210220T174502_036677_044F60_51BC
S1A_IW_GRDH_1SDV_20200827T022348_20200827T022413_034086_03F522_A17C
S1A_IW_GRDH_1SDV_20210328T174437_20210328T174502_037202_046199_9555
S1B_IW_GRDH_1SDV_20200917T151805_20200917T151830_023417_02C7B5_BCBB
S1B_IW_GRDH_1SDV_20200824T152621_20200824T152646_023067_02BCC0_1F9D
S1A_IW_GRDH_1SDV_20210430T155209_20210430T155234_037682_047231_FE15
S1A_IW_GRDH_1SDV_20200912T032519_20200912T032544_034320_03FD6B_6C6C
S1B_IW_GRDH_1SDV_20200807T041503_20200807T041528_022812_02B4BF_B5B6
S1A_IW_GRDH_1SDV_20210907T142530_20210907T142555_039577_04ADA3_0DDB
S1A_IW_GRDH_1SDV_20200805T144239_20200805T144304_033773_03EA39_1ED1

to reproduce,

git checkout oneformer-pl
docker build -t slickserver-pl
docker run -it --rm \
    -v $HOME/.aws:/root/.aws \
    -v "$(pwd)":/slickformer \
    -p 8888:8888 \
    -e AWS_PROFILE=devseed \
    --gpus all slickserver-pl

then connect to vscode with Remote Containers. the .vscode with a launch json config titled "make coco dataset no tile" should already exist. run that with the cli.py open to run the debugger

rbavery commented 1 year ago

:facepalm: we aren't sourcing the images from the open S1 bucket, we're sourcing them from the cerulean skytruth aws account. setting the aws profile correctly fixes this.