Datasets:

Modalities:
Image
Text
Formats:
parquet
Languages:
English
ArXiv:
Libraries:
Datasets
Dask
Zhong commited on
Commit
89f273f
·
verified ·
1 Parent(s): fc21cb1

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -94,7 +94,7 @@ Each entry in the dataset contains the following fields:
94
  - `boxed_image`: The original image overlaid with a red bounding box highlighting the target object.
95
  - `box_xmin`, `box_xmax`, `box_ymin`, `box_ymax`: The normalized bounding‑box coordinates.
96
  - `is_coco`: A binary flag (1 for COCO-class, 0 for non‑COCO).
97
- - `label_name`: The object’s category label (e.g., “muffin,” “giraffe”).
98
  - `co_occurrence`: The number of same‑class instances in the image (1 = no distractors; >1 = multiple).
99
  - `written_descriptions`: Three human‑typed referring expressions.
100
  - `spoken_descriptions`: Two human‑spoken expressions (transcribed and optionally corrected by annotators).
 
94
  - `boxed_image`: The original image overlaid with a red bounding box highlighting the target object.
95
  - `box_xmin`, `box_xmax`, `box_ymin`, `box_ymax`: The normalized bounding‑box coordinates.
96
  - `is_coco`: A binary flag (1 for COCO-class, 0 for non‑COCO).
97
+ - `label_name`: The object’s class label (e.g., “muffin,” “giraffe”).
98
  - `co_occurrence`: The number of same‑class instances in the image (1 = no distractors; >1 = multiple).
99
  - `written_descriptions`: Three human‑typed referring expressions.
100
  - `spoken_descriptions`: Two human‑spoken expressions (transcribed and optionally corrected by annotators).