search

ftbabi / LayersNet_ICCV2023

Apache License 2.0
38 stars 4 forks source link

stat.h5 missing #2

Closed Dolorousrtur closed 5 months ago

Dolorousrtur commented 8 months ago

Hi, thanks for publishing this amazing project!

I'd like to train a LayersNet model on the LAYERS dataset following your instructions.

I've downloaded the LAYERS dataset, unpacked it, and ran 

python tools/preprocessing_data.py configs/layersnet/base/ep1.py --work_dir *my_log_dir* --dataset test --type static

However, it asks for some stat.h5 file (this line of code) which seems to be missing from the published dataset.

Could you help me find it?

Many thanks in advance!

Ribosome-rbx commented 8 months ago

hi, I'm also facing the same problem. I really appreciate the incredible work you've put into the LayersNet project – it's truly impressive. Could you please provide clarification on the availability or location of the stat.h5 file or offer guidance on how to proceed in its absence? Your assistance would be immensely valuable.

Thank you very much for your time and consideration.

ForrestPi commented 7 months ago

@ftbabi Hi, hoping for help

ftbabi commented 5 months ago

Hi,

Sorry for the late reply.

"stat.h5" is a file including the mean and std for normalizing the input data. As indicated in "Dataset Preparation" Step 2, you may run the following command to generate required data:

python tools/preprocessing_data.py configs/layersnet/base/ep1.py --work_dir PATH/TO/LOG/DIR --dataset [train/val/test] --type dynamic

The "stat.h5" should be generated after this command. The related codes locate at L231 in LayersNet_ICCV2023layersnet/datasets/layers_dynamic_dataset.py.

Best, Yidi