Structure of PartImageNet dataset

[HashmatShadab]

I downloaded data PartImageNet_Seg, from https://github.com/tacju/partimagenet. After unzipping, the dataset structure is given below:

The link using gdwon doesn't download the dataset, it only downloads the json files for train, val and test.

So,

1. mkdir JPEGImages
2. mv train/* test/* val/* JPEGImages
   This will move all the images in the JPEGImages.

However, the below function in preparing dataset expects the data to have a subfolder in the JPEGImages:

def get_all_image_names(path):
    all_names = glob.glob(f"{path}/JPEGImages/*/*.JPEG")
    all_names = [name.split("/")[-1].split(".")[0] for name in all_names]
    return set(all_names)

Can you please provide more details regarding data preparation and let me know if I am making any mistake?