Upload README.md with huggingface_hub

README.md

@@ -1,104 +1,43 @@
 ---
-language:
-- en
-license: unknown
-task_categories:
-- image-segmentation
-- object-detection
-tags:
-- object-detection
-- bounding-boxes
-- relation-ordering
-- ADE20K
-- COCO
-pretty_name: ROPE
-size_categories:
-- 1K<n<10K
+dataset_info:
+  features:
+  - name: image
+    dtype: image
+  - name: label
+    dtype: string
+  - name: split_type
+    dtype: string
+  - name: objects
+    dtype: string
+  - name: object_set
+    dtype: string
+  - name: relations
+    dtype: string
+  - name: size
+    dtype: string
+  configs:
+  - config_name: default
+    data_files:
+    - split: train
+      path: train-*.parquet
+    - split: validation
+      path: validation-*.parquet
 ---
 
-# ROPE: Robust Object Pose Estimation through Relation Ordering
+# ROPE
 
-This dataset is a reformatted version of [sled-umich/ROPE](https://huggingface.co/datasets/sled-umich/ROPE),
-originally from the paper **"ROPE: Robust Object Pose Estimation through Relation Ordering"**.
+ROPE is an object detection benchmark combining images from ADE20K and COCO with bounding box annotations. 5,161 examples across 5 spatial relationship patterns (AAAAB, BAAAA, homogenous, heterogenous, mixed).
 
-## Dataset Description
+## Fields
 
-ROPE is an object detection benchmark that combines images from ADE20K and COCO datasets.
-Each image contains objects with bounding box annotations and object set information.
-The dataset evaluates models on their ability to handle different spatial relationship patterns
-between objects.
+| Field | Description |
+|-------|-------------|
+| image | Input image (with bounding box overlays) |
+| label | Source dataset: `ADE` or `COCO` |
+| split_type | Relationship pattern |
+| objects | JSON-encoded list of objects with bounding boxes |
+| object_set | JSON-encoded list of object names |
+| relations | JSON-encoded spatial relations |
+| size | JSON-encoded image dimensions |
 
-### Splits
-
-| Split       | Rows  |
-|-------------|-------|
-| train       | 2,587 |
-| validation  | 2,574 |
-
-### Split Types
-
-Each image belongs to one of five split types based on the spatial relationship pattern:
-
-| Split Type      | Train | Validation | Description                              |
-|-----------------|-------|------------|------------------------------------------|
-| AAAAB           | 168   | 334        | Five objects, one different from the rest |
-| BAAAA           | 168   | 334        | Five objects, first is different          |
-| homogenous      | 400   | 490        | Objects from same category               |
-| heterogenous    | 312   | 246        | Objects from different categories         |
-| mixed           | 1,539 | 1,170      | Mixed category composition               |
-
-### Labels (Source Dataset)
-
-| Source | Train | Validation |
-|--------|-------|------------|
-| ADE    | 1,379 | 1,322      |
-| COCO   | 1,208 | 1,252      |
-
-## Schema
-
-| Column      | Type                        | Description                                                    |
-|-------------|-----------------------------|----------------------------------------------------------------|
-| image       | struct\<bytes, path\>       | The image in HF image format (binary bytes + filename)         |
-| label       | string                      | Source dataset: "ADE" or "COCO"                                |
-| split_type  | string                      | Relationship pattern: "AAAAB", "BAAAA", "homogenous", "heterogenous", or "mixed" |
-| objects     | string (JSON)               | JSON-encoded list of objects with bounding boxes (name, object_id, bndbox) |
-| object_set  | string (JSON)               | JSON-encoded list of all object names present in the image     |
-| relations   | string (JSON)               | JSON-encoded list of spatial relations between objects         |
-| size        | string (JSON)               | JSON-encoded image dimensions (width, height, depth)           |
-
-### Example Objects Format
-
-```json
-[
-  {
-    "name": "person",
-    "object_id": "13",
-    "difficult": "0",
-    "bndbox": {"xmin": 244, "ymin": 0, "xmax": 290, "ymax": 123},
-    "bbox_number": 1
-  }
-]
-```
-
-## Differences from Original
-
-This version differs from the original `sled-umich/ROPE` dataset:
-
-1. **Annotations merged**: The original stored JSON annotation files separately from the image dataset.
-   This version embeds the annotations directly as columns.
-2. **Raw images removed**: The original included both "bbox" (with bounding box overlays) and "raw" (original)
-   versions of each image. This version only includes the bbox images.
-3. **Corrected labels**: The original imagefolder-based labels were all "ADE" due to directory structure.
-   This version uses the actual `data_source` field from annotations.
-4. **Unified schema**: All metadata is available in a single table rather than requiring separate file lookups.
-
-## Citation
-
-```bibtex
-@inproceedings{shan2024rope,
-  title={ROPE: Robust Object Pose Estimation through Relation Ordering},
-  author={Shan, Sijie and others},
-  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
-  year={2024}
-}
-```
+Converted from [sled-umich/ROPE](https://huggingface.co/datasets/sled-umich/ROPE).