Hello,
I hope this message finds you well.
I am currently attempting to use the action category labels from the THUMOS14 dataset to train the ProText model, aiming to obtain better text embeddings.
However, I noticed that THUMOS14 is not included as a built-in dataset in the Dassl.pytorch framework.
I encountered the following error:
Traceback (most recent call last):
File "train.py", line 257, in
main(args)
File "train.py", line 192, in main
trainer = build_trainer(cfg)
File "/root/ProText/Dassl.pytorch/dassl/engine/build.py", line 11, in build_trainer
return TRAINER_REGISTRY.get(cfg.TRAINER.NAME)(cfg)
File "/root/ProText/Dassl.pytorch/dassl/engine/trainer.py", line 324, in init
self.build_data_loader()
File "/root/ProText/Dassl.pytorch/dassl/engine/trainer.py", line 347, in build_data_loader
dm = DataManager(self.cfg)
File "/root/ProText/Dassl.pytorch/dassl/data/data_manager.py", line 79, in init
dataset = build_dataset(cfg)
File "/root/ProText/Dassl.pytorch/dassl/data/datasets/build.py", line 8, in build_dataset
check_availability(cfg.DATASET.NAME, avai_datasets)
File "/root/ProText/Dassl.pytorch/dassl/utils/tools.py", line 174, in check_availability
raise ValueError(
ValueError: The requested one is expected to belong to ['Digit5', 'VisDA17', 'CIFARSTL', 'Office31', 'DomainNet', 'OfficeHome', 'miniDomainNet', 'PACS', 'VLCS', 'FMoW', 'IWildCam', 'Camelyon17', 'CIFAR10C', 'CIFAR100C', 'DigitsDG', 'DigitSingle', 'OfficeHomeDG', 'CIFAR10', 'CIFAR100', 'SVHN', 'STL10', 'OxfordPets', 'OxfordFlowers', 'FGVCAircraft', 'DescribableTextures', 'EuroSAT', 'StanfordCars', 'Food101', 'SUN397', 'Caltech101', 'UCF101', 'ImageNet', 'ImageNetSketch', 'ImageNetV2', 'ImageNetA', 'ImageNetR'], but got [THUMOS14] (do you mean [STL10]?)
Given this, I would like to ask for your advice:
Would you recommend implementing a custom dataset module to integrate THUMOS14 with ProText?
Or, would you suggest any other approach based on your experience?
I would greatly appreciate any guidance you could provide.
Thank you very much for your time and support!
Best regards,
Shef, Shih
muzairkhattak
