Datasets:

Modalities:
Image
Text
Formats:
parquet
Languages:
English
ArXiv:
Libraries:
Datasets
Dask
Zhong commited on
Commit
01a1de9
·
verified ·
1 Parent(s): 1541b82

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +2 -1
README.md CHANGED
@@ -92,7 +92,8 @@ For token-level annotation of referring expressions, see the companion dataset [
92
  Each entry in the dataset contains the following fields:
93
 
94
  - `image`: The original image file.
95
- - `mask`: A binary segmentation mask isolating the target object.
 
96
  - `boxed_image`: The original image overlaid with a red bounding box highlighting the target object.
97
  - `box_xmin`, `box_xmax`, `box_ymin`, `box_ymax`: The normalized bounding‑box coordinates.
98
  - `is_coco`: A binary flag (1 for COCO-class, 0 for non‑COCO).
 
92
  Each entry in the dataset contains the following fields:
93
 
94
  - `image`: The original image file.
95
+ - `mask`: A binary segmentation mask isolating the target object.
96
+ - `related_masks`: A list of other masks of the class in the image, will be empty for a `single_precence` entry.
97
  - `boxed_image`: The original image overlaid with a red bounding box highlighting the target object.
98
  - `box_xmin`, `box_xmax`, `box_ymin`, `box_ymax`: The normalized bounding‑box coordinates.
99
  - `is_coco`: A binary flag (1 for COCO-class, 0 for non‑COCO).