Zhong
commited on
Update README.md
Browse files
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
|
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).
|