Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
annotations_creators:
|
| 3 |
- expert-generated
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pretty_name: DOCCI (preprocessed subset for Mini-LLaVA)
|
| 3 |
+
license: cc-by-4.0
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
This dataset repackages a subset of **[google/docci]** images and descriptions
|
| 7 |
+
for Mini-LLaVA training/inference. We only reorganized files and converted
|
| 8 |
+
metadata to `data/docci_converted.json`. **Original content and copyright remain
|
| 9 |
+
with the DOCCI authors.**
|
| 10 |
+
|
| 11 |
+
**Attribution:** Onoe et al., *DOCCI: Descriptions of Connected and Contrasting Images*, ECCV 2024.
|
| 12 |
+
Original dataset: google/docci (CC BY 4.0).
|
| 13 |
+
|
| 14 |
+
**Changes from the original:** file renaming, directory layout, and metadata JSON conversion.
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
Original README.md is shown below.
|
| 18 |
+
|
| 19 |
---
|
| 20 |
annotations_creators:
|
| 21 |
- expert-generated
|