can you provide the details of dataset preparation?

[gouchaonijiao]

Can you provide the details of dataset preparation? Since I can not understand the meaning of the 15 attributes in "shape ann file". Thanks!

[yumingj]

Hi The definition of the the attributes are defined here: https://github.com/yumingj/Text2Human/blob/b4ffaf6cd82b195f258c8608455374147c188f99/utils/language_utils.py#L85

They are a little bit different from the original dataset annotations.

[gouchaonijiao]

thanks for your reply and your excellent work too!!

[gouchaonijiao]

Hi The definition of the the attributes are defined here:

https://github.com/yumingj/Text2Human/blob/b4ffaf6cd82b195f258c8608455374147c188f99/utils/language_utils.py#L85

They are a little bit different from the original dataset annotations.

hello! I have some quetions about the following code " for text_id, text in enumerate(parsed_texts): user_embeddings = model.encode(text) if ('man' in text) and (text_id == 0): human_attr[0] = 0 human_attr[1] = 0

    if ('woman' in text or 'lady' in text) and (text_id == 0):
        human_attr[0] = 1
        human_attr[1] = 2"

according to the code,why there are some human_attr's value are [0,3], like MEN-Denim-id_00005447-01_4_full.png 0 3 3 3 0 2 4 3 0 1 1 0 0 1 0

[yumingj]

I enforce the attributes value using the if-clause for the better inference results, as it is hard to generate man with long hair due to the dataset bias.

[gouchaonijiao]

Thanks a lot for your explanation!