training_set and testing_set of tvsum

[pcshih]

As mention in the paper, the training and testing set should be 80% and 20%. But in https://github.com/weirme/Video_Summary_using_FCSN/blob/96b40851b7805afd1f1fc69f2beb5143d5727b4e/data_loader.py#L25

should it be train_dataset, test_dataset = torch.utils.data.random_split(dataset, [int(len(dataset)*0.8), int(len(dataset)*0.2)])?

Thank you.

[pcshih]

Could you give me the Summe dataset (downsampled to 320 frames per video) preprocessed by make_dataset.py? I would like to try the evaluation.

[weirme]

Could you give me the Summe dataset (downsampled to 320 frames per video) preprocessed by make_dataset.py? I would like to try the evaluation.

Sorry, the hdf5 file is saved on a server in my school but now I am at home... You may make it using make_dataset.py.

[pcshih]

Is the performance far from the paper mentioned(SumME)?

[weirme]

Is the performance far from the paper mentioned(SumME)?

Yes

[pcshih]

I cannot run the code. What is the input format of video? i.e. Air_Force_One.mp4

[weirme]

I cannot run the code. What is the input format of video? i.e. Air_Force_One.mp4

You may run python make_dataset.py --video_dir {arg1} --h5_path {arg2} --vsumm_data {arg3}

and here are some descriptions about 3 args: https://github.com/weirme/Video_Summary_using_FCSN/blob/0895cccbb2a488369b1bfc7d2c087b3050250898/make_dataset.py#L15-L17

in case of SumMe, video_dir may like SumMe_root/videos.

[pcshih]

I also got bad performance in summe dataset.

[pcshih]

How was your performance in summe dataset?

[weirme]

Almost as bad as yours :(

[habib1402]

Is there any solution to the bad performance of SumMe dataset? Please someone guide me accordingly.