ValueError: invalid literal for int() with base 10: 'drumming'

[Leozyc-waseda]

Thank you for your excellent work.
I want to test the performance of the model on K400 using SlowFast, and I have set up the environment and dataset according to the requirements of install.md and dataset.md. The generated test.csv file is as shown below.

I used this command, but got this error. I can't understand, maybe someone has encountered the same error?

(slowfast) pana-tat@panatat-4CXE:~/work/video_classification/SlowFast$ python tools/run_net.py   --cfg /home/pana-tat/work/video_classification/SlowFast/configs/Kinetics/SLOWFAST_8x8_R50.yaml   DATA.PATH_TO_DATA_DIR '/home/pana-tat/work/video_classification/SlowFast/test.csv'     TEST.CHECKPOINT_FILE_PATH '/home/pana-tat/work/video_classification/PySlowFast_Model_Zoo/SLOWFAST_8x8_R50.pkl' 
...
[05/09 00:35:49][INFO] checkpoint.py:  701: Unknown way of loading checkpoint. Using with random initialization, only for debugging.
[05/09 00:35:49][INFO] kinetics.py:   93: Constructing Kinetics test...
Traceback (most recent call last):
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 52, in <module>
    main()
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 37, in main
    launch_job(cfg=cfg, init_method=args.init_method, func=test)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/utils/misc.py", line 430, in launch_job
    func(cfg=cfg)
  File "/home/pana-tat/work/video_classification/SlowFast/tools/test_net.py", line 218, in test
    test_loader = loader.construct_loader(cfg, "test")
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/loader.py", line 114, in construct_loader
    dataset = build_dataset(dataset_name, cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/build.py", line 31, in build_dataset
    return DATASET_REGISTRY.get(name)(cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 94, in __init__
    self._construct_loader()
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 150, in _construct_loader
    self._labels.append(int(label))
ValueError: invalid literal for int() with base 10: 'drumming'

[XinyangHan]

Hi! I encountered this as well!! May I ask how you solved it? Looks like the labels should be integers, but I am confused how to construct these labels.

[42Viva]

Thank you for your excellent work. I want to test the performance of the model on K400 using SlowFast, and I have set up the environment and dataset according to the requirements of install.md and dataset.md. The generated test.csv file is as shown below.

I used this command, but got this error. I can't understand, maybe someone has encountered the same error?

(slowfast) pana-tat@panatat-4CXE:~/work/video_classification/SlowFast$ python tools/run_net.py   --cfg /home/pana-tat/work/video_classification/SlowFast/configs/Kinetics/SLOWFAST_8x8_R50.yaml   DATA.PATH_TO_DATA_DIR '/home/pana-tat/work/video_classification/SlowFast/test.csv'     TEST.CHECKPOINT_FILE_PATH '/home/pana-tat/work/video_classification/PySlowFast_Model_Zoo/SLOWFAST_8x8_R50.pkl' 
...
[05/09 00:35:49][INFO] checkpoint.py:  701: Unknown way of loading checkpoint. Using with random initialization, only for debugging.
[05/09 00:35:49][INFO] kinetics.py:   93: Constructing Kinetics test...
Traceback (most recent call last):
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 52, in <module>
    main()
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 37, in main
    launch_job(cfg=cfg, init_method=args.init_method, func=test)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/utils/misc.py", line 430, in launch_job
    func(cfg=cfg)
  File "/home/pana-tat/work/video_classification/SlowFast/tools/test_net.py", line 218, in test
    test_loader = loader.construct_loader(cfg, "test")
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/loader.py", line 114, in construct_loader
    dataset = build_dataset(dataset_name, cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/build.py", line 31, in build_dataset
    return DATASET_REGISTRY.get(name)(cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 94, in __init__
    self._construct_loader()
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 150, in _construct_loader
    self._labels.append(int(label))
ValueError: invalid literal for int() with base 10: 'drumming'

May i ask how did u solve that? i guess we should change the 'int(label)' into '(label)'?

[burgosmad]

Thank you for your excellent work. I want to test the performance of the model on K400 using SlowFast, and I have set up the environment and dataset according to the requirements of install.md and dataset.md. The generated test.csv file is as shown below.
I used this command, but got this error. I can't understand, maybe someone has encountered the same error?

(slowfast) pana-tat@panatat-4CXE:~/work/video_classification/SlowFast$ python tools/run_net.py   --cfg /home/pana-tat/work/video_classification/SlowFast/configs/Kinetics/SLOWFAST_8x8_R50.yaml   DATA.PATH_TO_DATA_DIR '/home/pana-tat/work/video_classification/SlowFast/test.csv'     TEST.CHECKPOINT_FILE_PATH '/home/pana-tat/work/video_classification/PySlowFast_Model_Zoo/SLOWFAST_8x8_R50.pkl' 
...
[05/09 00:35:49][INFO] checkpoint.py:  701: Unknown way of loading checkpoint. Using with random initialization, only for debugging.
[05/09 00:35:49][INFO] kinetics.py:   93: Constructing Kinetics test...
Traceback (most recent call last):
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 52, in <module>
    main()
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 37, in main
    launch_job(cfg=cfg, init_method=args.init_method, func=test)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/utils/misc.py", line 430, in launch_job
    func(cfg=cfg)
  File "/home/pana-tat/work/video_classification/SlowFast/tools/test_net.py", line 218, in test
    test_loader = loader.construct_loader(cfg, "test")
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/loader.py", line 114, in construct_loader
    dataset = build_dataset(dataset_name, cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/build.py", line 31, in build_dataset
    return DATASET_REGISTRY.get(name)(cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 94, in __init__
    self._construct_loader()
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 150, in _construct_loader
    self._labels.append(int(label))
ValueError: invalid literal for int() with base 10: 'drumming'

May i ask how did u solve that? i guess we should change the 'int(label)' into '(label)'?

As I found that val.csv has labels with no label, it returns the error ValueError: invalid literal for int() with base 10: '\n'

I tried to change the 'int(label)' into '(label)' but no, if I do that the problem becomes to IndexError: only integers, slices (:), ellipsis (...), numpy.newaxis (None) and integer or boolean arrays are valid indices

[LSKhappychild]

Any update on this?

btw, where did you get the label csv files?

[mumairhassan]

Hi,

I am also facing the same issue. Is there any solution to this?

File "/home/muhammad.u.hassan/code/SlowFast/slowfast/datasets/kinetics.py", line 143, in _construct_loader
    self._labels.append(int(label))
ValueError: invalid literal for int() with base 10: 'drumming,013SMb0SX8I,20,30,test,0'

[pepper725]

Hi,

I am also facing the same issue. Is there any solution to this?

File "/home/muhammad.u.hassan/code/SlowFast/slowfast/datasets/kinetics.py", line 143, in _construct_loader
    self._labels.append(int(label))
ValueError: invalid literal for int() with base 10: 'drumming,013SMb0SX8I,20,30,test,0'

Have you solved the problem?

[tina908]

Thank you for your excellent work. I want to test the performance of the model on K400 using SlowFast, and I have set up the environment and dataset according to the requirements of install.md and dataset.md. The generated test.csv file is as shown below.

I used this command, but got this error. I can't understand, maybe someone has encountered the same error?

(slowfast) pana-tat@panatat-4CXE:~/work/video_classification/SlowFast$ python tools/run_net.py   --cfg /home/pana-tat/work/video_classification/SlowFast/configs/Kinetics/SLOWFAST_8x8_R50.yaml   DATA.PATH_TO_DATA_DIR '/home/pana-tat/work/video_classification/SlowFast/test.csv'     TEST.CHECKPOINT_FILE_PATH '/home/pana-tat/work/video_classification/PySlowFast_Model_Zoo/SLOWFAST_8x8_R50.pkl' 
...
[05/09 00:35:49][INFO] checkpoint.py:  701: Unknown way of loading checkpoint. Using with random initialization, only for debugging.
[05/09 00:35:49][INFO] kinetics.py:   93: Constructing Kinetics test...
Traceback (most recent call last):
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 52, in <module>
    main()
  File "/home/pana-tat/work/video_classification/SlowFast/tools/run_net.py", line 37, in main
    launch_job(cfg=cfg, init_method=args.init_method, func=test)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/utils/misc.py", line 430, in launch_job
    func(cfg=cfg)
  File "/home/pana-tat/work/video_classification/SlowFast/tools/test_net.py", line 218, in test
    test_loader = loader.construct_loader(cfg, "test")
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/loader.py", line 114, in construct_loader
    dataset = build_dataset(dataset_name, cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/build.py", line 31, in build_dataset
    return DATASET_REGISTRY.get(name)(cfg, split)
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 94, in __init__
    self._construct_loader()
  File "/home/pana-tat/work/video_classification/SlowFast/slowfast/datasets/kinetics.py", line 150, in _construct_loader
    self._labels.append(int(label))
ValueError: invalid literal for int() with base 10: 'drumming'

May I ask how did you solve that?

[LSKhappychild]

@tina908

I went thru the same problem, and found that there exists an extra csv file for mapping the literal label values into integer value (e.g. drumming -> 15). You can find it by googling some repositories regarding k400 dataset.

As I remember,

https://gist.github.com/willprice/f19da185c9c5f32847134b87c1960769

this one worked fine with me.

[tina908]

@LSKhappychild

Thank you very much for your response.
Could you please provide a more detailed explanation? I am a beginner and would greatly appreciate your guidance.

[LSKhappychild]

@tina908

As the log says, casting 'label', which is a string like 'drumming', is causing the error. So converting each action label (drumming) into integer number is needed.
Follow the link I attached, then as you can see in the first row, "abseiling" is converted to integer id 0.
This kind of conversion will be needed.