DATASET.md

Preparing Datasets for HyperSeg

Expected dataset structure for COCO:

coco/
  annotations/
    instances_{train,val}2017.json
    panoptic_{train,val}2017.json
  {train,val}2017/
    # image files that are mentioned in the corresponding json
  train2014/
    # image files for RefCOCO
  panoptic_{train,val}2017/  # png annotations
  panoptic_semseg_{train,val}2017/  # generated by the script mentioned in Mask2Former

Expected dataset structure for RES:

RES/
    refcoco/
        refcoco_val.json
        refcoco_testA.json
        refcoco_testB.json
    refcoco+/
        refcoco+_val.json
        refcoco+_testA.json
        refcoco+_testB.json
    refcocog/
        refcocog_val.json
        refcocog_test.json

Expected dataset structure for ReasonSeg:

ReasonSeg/
    train/
        image_1.jpg, image_1.json
        image_2.jpg, image_2.json
    val/
        image_1.jpg, image_1.json
        image_2.jpg, image_2.json

Expected dataset structure for ReVOS:

ReVOS/
    JPEGImages
        <video1  >
        <video2  >
        <video...>
    mask_dict.json
    mask_dict_foreground.json 
    meta_expressions_train_.json
    meta_expressions_valid_.json