gorkaydemir / SOLV

Official implementation of the NeurIPS 2023 paper "Self-supervised Object-Centric Learning for Videos"
22 stars 0 forks source link

YTVIS 2019 annotations #1

Open rm-wu opened 4 months ago

rm-wu commented 4 months ago

Hi, thank you for the great work! I wanted to replicate the experiments on ytvis 2019. I downloaded the files from here. However when training your model using the commands in the README.md I get an error during the evaluation phase because the code tries to load them from a folder called Annotations. Could you provide how did you download the dataset and how the dataset structure is organised?

Thanks

gorkaydemir commented 1 month ago

Hi, The dataset you downloaded is correct. Below is the expected structure that the code requires:

/path/to/root
│ 
├── train
│   ├── Annotations
│   │   ├── 003234408d (train - video 1)
│   │   │   ├── 00000.png
│   │   │   ├── 00010.png
│   │   │   └── ...
│   │   │ 
│   │   ├── 0043f083b5 (train - video 2)
│   │   │   ├── 00000.png
│   │   │   ├── 00010.png
│   │   │   └── ...
│   │   │ 
│   │   └──  ....
│   │  
│   ├── JPEGImages
│   │   ├── 003234408d (train - video 1)
│   │   │   ├── 00000.jpg
│   │   │   ├── 00010.jpg
│   │   │   └── ...
│   │   │ 
│   │   ├── 0043f083b5 (train - video 2)
│   │   │   ├──  00000.jpg
│   │   │   ├── 00010. jpg
│   │   │   └── ...
│   │   │ 
│   │   └──  ....
│   │   
│   ├── meta.json
│   └── train.json
│ 
└── val
    ├── JPEGImages
    │   ├── 0062f687f1 (val - video 1)
    │   │   ├── 00000.jpg
    │   │   ├── 00010.jpg
    │   │   └── ...
    │   │   
    │   ├── 0c04834d61 (val - video 2)
    │   │   ├── 00000.jpg
    │   │   ├── 00010.jpg
    │   │   └── ...
    │   │   
    │   └──  ....
    │  
    ├── meta.json
    └── valid.json