ImageNet Directory: /datasets01/imagenet_full_size/061417/

vision
1

I am trying to run the code in this repository https://github.com/facebookresearch/deepcluster. I am using windows 10. When I try to run main.sh, nothing happens. I opened the file and there is this line DIR="/datasets01/imagenet_full_size/061417/train". I know that they are using ImageNet but what I don’t understand is that whether the dataset is downloaded by the code and this directory is created or whether I need to download it myself? I checked other codes and they also used the same path. I appreciate any help and guidance regarding this and also running the code in general.

Thanks

2

same question，have you solved this now? :smiling_face_with_tear:

3

Hello
I found out that you can use it without the [/datasets01/imagenet_full_size/061417/] dataset
just create your own directory : /dataset/imagenet_full_size/061417/
then add 2 folders (train and test) from any dataset on kaggle

.
└── 061417
    β”œβ”€β”€ test
    β”‚   β”œβ”€β”€ cats
    β”‚   β”‚   β”œβ”€β”€ cat_106.jpg
    β”‚   β”‚   β”œβ”€β”€ cat_109.jpg
    β”‚   β”‚   β”œβ”€β”€ .........
    β”‚   └── dogs
    β”‚       β”œβ”€β”€ dog_114.jpg
    β”‚       β”œβ”€β”€ dog_123.jpg
    β”‚       β”œβ”€β”€ ..........

    └── train
        β”œβ”€β”€ cats
        β”‚   β”œβ”€β”€ cat_100.jpg
        β”‚   β”œβ”€β”€ ........
        └── dogs
            β”œβ”€β”€ dog_0.jpg
            β”œβ”€β”€ .........

check this out IMAGEFOLDER - Pytorch configuration

1 Like