Update README.md

README.md

@@ -9,4 +9,59 @@ language:
 - en
 tags:
 - art
+---
+
+# AmbiBench Metadata
+
+This dataset provides the **metadata** used in the AmbiBench benchmark.  
+Each row corresponds to an ambiguous image or video sample paired with a question–answer pair, derived from the benchmark design.
+
+---
+
+## Columns
+
+### `question`
+- The natural-language question posed to the model.  
+- Questions are aligned with AmbiBench’s four task types:
+  - **Open-ended**: ask for all possible interpretations (e.g., “What hidden objects can you see in the image?”).  
+  - **Multiple-choice**: ask to select all valid interpretations from candidates.  
+  - **Ambiguous localization**: ask to provide bounding boxes for ambiguous regions.  
+  - **Local region description**: ask to describe multiple interpretations of a highlighted region.
+
+### `answer`
+- The reference answer to the `question`.  
+- May include:
+  - A single object name (`"snake"`),  
+  - A semicolon-separated pair for bistable/multi-scene (`"duck; rabbit"`),  
+  - Or multiple objects/scenes (`"bear, deer, wolf, bird"`).
+
+### `file_name`
+- Cleaned filename of the associated image or video.  
+- Any inline bounding box tags in the original filename (e.g.  
+  `2120_netimg_hidden_face_illusion_[1152,807,1392,1263].jpg`)  
+  are removed, and the coordinates are stored separately in `bbox`.  
+- ⚠️ **Special case: `motion_video` samples** — the stored file names end with `.png` as placeholders.  
+  For actual usage, replace `.png` with `.mp4` to access the corresponding video file.
+
+### `bbox`
+- A list of bounding box coordinates `[x1, y1, x2, y2]` if required by the question.  
+- Empty list (`[]`) if no bounding box is provided.  
+
+### `type`
+- Combined label = **Ambiguity category** (*from the AmbiBench taxonomy*) + **Task type**.  
+- Categories include:  
+  *Camouflaged, Bistable, Hybrid, Color, Multi-view, Geometric, Motion, Multi-scene, Mixed*.  
+- Task types include: *open-ended, localization(bbox), region*.  
+- Example values:  
+  - `"camouflaged_open"`  
+  - `"mix_bbox"`  
+  - `"mix_highlighted"`
+
+---
+
+## Notes
+- All entries include consistent fields: `question`, `answer`, `file_name`, `bbox`, `type`.  
+- `motion_video` entries must be linked to the corresponding `.mp4` files (replace `.png` extension).  
+- This metadata is suitable for training and evaluating vision–language models on ambiguous image understanding.
+
 ---