Skip to content

Latest commit

 

History

History
35 lines (28 loc) · 905 Bytes

data.md

File metadata and controls

35 lines (28 loc) · 905 Bytes

Data

Our experiments primarily use the ImageNet-1k dataset. Please organize the dataset in the following structure:

data/imagenet/
├── annotations
│   ├── train.json
|   └── val.json
├── train
│   ├── n1440764
│   │   ├── 18.JPEG
│   │   └── ...
|   └── ...
├── val
│   ├── n1440764
│   │   ├── 293.JPEG
│   │   └── ...
|   └── ...
└── synsets.json

Both train.json and val.json have the following structure:

[{"image":"12925.JPEG","synset_id":449},...]

synsets.json maps synset IDs to their corresponding information:

{"1":{"WNID":"n02119789","words":"kit fox, Vulpes macrotis",...},...}

For detailed instructions on downloading and preparing the dataset, refer to https://toddai.readthedocs.io/en/latest/data/imagenet.html#imagenet-1k