search

microsoft / scene_graph_benchmark

image scene graph generation benchmark
MIT License
382 stars 86 forks source link

Cannot run VinVL feature extraction command #37

Open lukerm opened 3 years ago

lukerm commented 3 years ago

I ran the following command (from the README):

python tools/test_sg_net.py --config-file sgg_configs/vgattr/vinvl_x152c4.yaml TEST.IMS_PER_BATCH 2 MODEL.WEIGHT models/vinvl/vinvl_vg_x152c4.pth MODEL.ROI_HEADS.NMS_FILTER 1 MODEL.ROI_HEADS.SCORE_THRESH 0.2 DATA_DIR "../maskrcnn-benchmark-1/datasets1" TEST.IGNORE_BOX_REGRESSION True MODEL.ATTRIBUTE_ON True

I think the DATA_DIR is misconfigured because I get the following error (below). Where is ../maskrcnn-benchmark-1/datasets1 from? Or the file visualgenome/test_vgoi6_clipped.yaml, which I think it's looking for?

This is the AssertionError:

Traceback (most recent call last):
  File "tools/test_sg_net.py", line 197, in <module>
    main()
  File "tools/test_sg_net.py", line 193, in main
    run_test(cfg, model, args.distributed, model_name)
  File "tools/test_sg_net.py", line 55, in run_test
    data_loaders_val = make_data_loader(cfg, is_train=False, is_distributed=distributed)
  File "/scene_graph_benchmark/maskrcnn_benchmark/data/build.py", line 170, in make_data_loader
    datasets = build_dataset(cfg, transforms, DatasetCatalog, is_train or is_for_period)
  File "/scene_graph_benchmark/maskrcnn_benchmark/data/build.py", line 45, in build_dataset
    cfg, dataset_name, factory_name, is_train
  File "/scene_graph_benchmark/maskrcnn_benchmark/data/datasets/utils/config_args.py", line 7, in config_tsv_dataset_args
    assert op.isfile(full_yaml_file)
AssertionError

Thanks in advance!

joeyy5588 commented 3 years ago

You can download the dataset following the instruction in the Perform training section in readme. The path to your yaml file would be DATA_DIR followed by DATASETS.TEST in your testing config ("../maskrcnn-benchmark-1/datasets1" + "visualgenome/test_vgoi6_clipped.yaml") for example.

lukerm commented 3 years ago

I have downloaded the data, but have not found test_vgoi6_clipped.yaml. Here are the steps I took:

$ ./azcopy copy 'https://penzhanwu2.blob.core.windows.net/sgg/sgg_benchmark/datasets/visualgenome' data --recursive
INFO: Scanning...
INFO: Any empty folders will not be processed, because source and/or destination doesn't have full folder support

Job 1b034082-73a8-0a41-67de-1ae538b27c87 has started

The downloaded YAML files does not include test_vgoi6_clipped.yaml

$ ls -la data/visualgenome/*.yaml 
-rw-r--r-- 1 ubuntu ubuntu 169 Aug  2 13:06 data/visualgenome/test_danfeiX_relation.yaml
-rw-r--r-- 1 ubuntu ubuntu 170 Aug  2 13:06 data/visualgenome/train_danfeiX_relation_nm.yaml

For your reference, here is everything that was downloaded:

$ ls -la data/visualgenome/                        
total 20264496
drwxrwxr-x 2 ubuntu ubuntu        4096 Aug  2 13:11 .
drwxrwxr-x 3 ubuntu ubuntu        4096 Aug  2 13:06 ..
-rw-r--r-- 1 ubuntu ubuntu      835049 Aug  2 13:06 hw.lineidx
-rw-r--r-- 1 ubuntu ubuntu     4108316 Aug  2 13:06 hw.tsv
-rw-r--r-- 1 ubuntu ubuntu     1241364 Aug  2 13:06 img.lineidx
-rw-r--r-- 1 ubuntu ubuntu 20439590250 Aug  2 13:11 img.tsv
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_clipped.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_clipped.new.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu      997885 Aug  2 13:06 label_danfeiX_clipped.new.lineidx
-rw-r--r-- 1 ubuntu ubuntu   145570762 Aug  2 13:06 label_danfeiX_clipped.new.tsv
-rw-r--r-- 1 ubuntu ubuntu      165464 Aug  2 13:06 label_danfeiX_clipped.relation.test.linelist.tsv
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_overlap.new.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu      996497 Aug  2 13:06 label_danfeiX_overlap.new.lineidx
-rw-r--r-- 1 ubuntu ubuntu   142911783 Aug  2 13:07 label_danfeiX_overlap.new.tsv
-rw-r--r-- 1 ubuntu ubuntu      356521 Aug  2 13:06 label_danfeiX_overlap.relation.train.linelist.tsv
-rw-r--r-- 1 ubuntu ubuntu         169 Aug  2 13:06 test_danfeiX_relation.yaml
-rw-r--r-- 1 ubuntu ubuntu         170 Aug  2 13:06 train_danfeiX_relation_nm.yaml
-rw-r--r-- 1 ubuntu ubuntu       28453 Aug  2 13:06 VG-SGG-dicts-danfeiX-clipped.json
-rw-r--r-- 1 ubuntu ubuntu       28453 Aug  2 13:06 VG-SGG-dicts-danfeiX-overlap.json

Is there somewhere else I should download it from?

linxi1158 commented 3 years ago

You can download the dataset following the instruction in the Perform training section in readme. The path to your yaml file would be DATA_DIR followed by DATASETS.TEST in your testing config ("../maskrcnn-benchmark-1/datasets1" + "visualgenome/test_vgoi6_clipped.yaml") for example.

I have downloaded the data, but have not found test_vgoi6_clipped.yaml. Here are the steps I took:

$ ./azcopy copy 'https://penzhanwu2.blob.core.windows.net/sgg/sgg_benchmark/datasets/visualgenome' data --recursive
INFO: Scanning...
INFO: Any empty folders will not be processed, because source and/or destination doesn't have full folder support

Job 1b034082-73a8-0a41-67de-1ae538b27c87 has started

The downloaded YAML files does not include test_vgoi6_clipped.yaml

$ ls -la data/visualgenome/*.yaml 
-rw-r--r-- 1 ubuntu ubuntu 169 Aug  2 13:06 data/visualgenome/test_danfeiX_relation.yaml
-rw-r--r-- 1 ubuntu ubuntu 170 Aug  2 13:06 data/visualgenome/train_danfeiX_relation_nm.yaml

For your reference, here is everything that was downloaded:

$ ls -la data/visualgenome/                        
total 20264496
drwxrwxr-x 2 ubuntu ubuntu        4096 Aug  2 13:11 .
drwxrwxr-x 3 ubuntu ubuntu        4096 Aug  2 13:06 ..
-rw-r--r-- 1 ubuntu ubuntu      835049 Aug  2 13:06 hw.lineidx
-rw-r--r-- 1 ubuntu ubuntu     4108316 Aug  2 13:06 hw.tsv
-rw-r--r-- 1 ubuntu ubuntu     1241364 Aug  2 13:06 img.lineidx
-rw-r--r-- 1 ubuntu ubuntu 20439590250 Aug  2 13:11 img.tsv
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_clipped.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_clipped.new.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu      997885 Aug  2 13:06 label_danfeiX_clipped.new.lineidx
-rw-r--r-- 1 ubuntu ubuntu   145570762 Aug  2 13:06 label_danfeiX_clipped.new.tsv
-rw-r--r-- 1 ubuntu ubuntu      165464 Aug  2 13:06 label_danfeiX_clipped.relation.test.linelist.tsv
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_overlap.new.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu      996497 Aug  2 13:06 label_danfeiX_overlap.new.lineidx
-rw-r--r-- 1 ubuntu ubuntu   142911783 Aug  2 13:07 label_danfeiX_overlap.new.tsv
-rw-r--r-- 1 ubuntu ubuntu      356521 Aug  2 13:06 label_danfeiX_overlap.relation.train.linelist.tsv
-rw-r--r-- 1 ubuntu ubuntu         169 Aug  2 13:06 test_danfeiX_relation.yaml
-rw-r--r-- 1 ubuntu ubuntu         170 Aug  2 13:06 train_danfeiX_relation_nm.yaml
-rw-r--r-- 1 ubuntu ubuntu       28453 Aug  2 13:06 VG-SGG-dicts-danfeiX-clipped.json
-rw-r--r-- 1 ubuntu ubuntu       28453 Aug  2 13:06 VG-SGG-dicts-danfeiX-overlap.json

Is there somewhere else I should download it from?

I have the same problem

hikaru-nara commented 3 years ago

Same problem here

cwkeam commented 2 years ago

Same problem here

cwkeam commented 2 years ago

We'll definitely need the proper inputs from the actual developers of this package, but seeing that this test_vgoi_clipped.yaml file probably serves the same function, one way to see the code run is by just doing

mv test_danfeiX_relation.yaml test_vgoi_clipped.yaml thought it made some sense considering that the purpose of this yaml file is just to point the code to the dataset files, and seems like all the dataset files in this visual genome are these "danfeiX" files.

linxi1158 commented 2 years ago

We'll definitely need the proper inputs from the actual developers of this package, but seeing that this test_vgoi_clipped.yaml file probably serves the same function, one way to see the code run is by just doing

mv test_danfeiX_relation.yaml test_vgoi_clipped.yaml thought it made some sense considering that the purpose of this yaml file is just to point the code to the dataset files, and seems like all the dataset files in this visual genome are these "danfeiX" files.

ok, thanks

via815via commented 2 years ago

We'll definitely need the proper inputs from the actual developers of this package, but seeing that this test_vgoi_clipped.yaml file probably serves the same function, one way to see the code run is by just doing mv test_danfeiX_relation.yaml test_vgoi_clipped.yaml thought it made some sense considering that the purpose of this yaml file is just to point the code to the dataset files, and seems like all the dataset files in this visual genome are these "danfeiX" files.

ok, thanks

Do you solve this problem? Can you tell me? Thanks very much!

Lxy811 commented 7 months ago

I have downloaded the data, but have not found test_vgoi6_clipped.yaml. Here are the steps I took:

$ ./azcopy copy 'https://penzhanwu2.blob.core.windows.net/sgg/sgg_benchmark/datasets/visualgenome' data --recursive
INFO: Scanning...
INFO: Any empty folders will not be processed, because source and/or destination doesn't have full folder support

Job 1b034082-73a8-0a41-67de-1ae538b27c87 has started

The downloaded YAML files does not include test_vgoi6_clipped.yaml

$ ls -la data/visualgenome/*.yaml 
-rw-r--r-- 1 ubuntu ubuntu 169 Aug  2 13:06 data/visualgenome/test_danfeiX_relation.yaml
-rw-r--r-- 1 ubuntu ubuntu 170 Aug  2 13:06 data/visualgenome/train_danfeiX_relation_nm.yaml

For your reference, here is everything that was downloaded:

$ ls -la data/visualgenome/                        
total 20264496
drwxrwxr-x 2 ubuntu ubuntu        4096 Aug  2 13:11 .
drwxrwxr-x 3 ubuntu ubuntu        4096 Aug  2 13:06 ..
-rw-r--r-- 1 ubuntu ubuntu      835049 Aug  2 13:06 hw.lineidx
-rw-r--r-- 1 ubuntu ubuntu     4108316 Aug  2 13:06 hw.tsv
-rw-r--r-- 1 ubuntu ubuntu     1241364 Aug  2 13:06 img.lineidx
-rw-r--r-- 1 ubuntu ubuntu 20439590250 Aug  2 13:11 img.tsv
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_clipped.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_clipped.new.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu      997885 Aug  2 13:06 label_danfeiX_clipped.new.lineidx
-rw-r--r-- 1 ubuntu ubuntu   145570762 Aug  2 13:06 label_danfeiX_clipped.new.tsv
-rw-r--r-- 1 ubuntu ubuntu      165464 Aug  2 13:06 label_danfeiX_clipped.relation.test.linelist.tsv
-rw-r--r-- 1 ubuntu ubuntu     4651532 Aug  2 13:06 label_danfeiX_overlap.new.freq_prior.npy
-rw-r--r-- 1 ubuntu ubuntu      996497 Aug  2 13:06 label_danfeiX_overlap.new.lineidx
-rw-r--r-- 1 ubuntu ubuntu   142911783 Aug  2 13:07 label_danfeiX_overlap.new.tsv
-rw-r--r-- 1 ubuntu ubuntu      356521 Aug  2 13:06 label_danfeiX_overlap.relation.train.linelist.tsv
-rw-r--r-- 1 ubuntu ubuntu         169 Aug  2 13:06 test_danfeiX_relation.yaml
-rw-r--r-- 1 ubuntu ubuntu         170 Aug  2 13:06 train_danfeiX_relation_nm.yaml
-rw-r--r-- 1 ubuntu ubuntu       28453 Aug  2 13:06 VG-SGG-dicts-danfeiX-clipped.json
-rw-r--r-- 1 ubuntu ubuntu       28453 Aug  2 13:06 VG-SGG-dicts-danfeiX-overlap.json

Is there somewhere else I should download it from?

Can you share the downloaded data?