Datasets:

maelic
/

OV-SGG-Bench

---
license: other
license_name: mixed-upstream-licenses
task_categories:
  - object-detection
language:
  - en
tags:
  - scene-graph-generation
  - visual-relationship-detection
  - open-vocabulary
  - benchmark
  - evaluation
pretty_name: OV-SGG-Bench
size_categories:
  - 10K<n<100K
---
# OV-SGG-Bench: evaluation packs for open-vocabulary relation prediction

The six evaluation sources of **OV-SGG-Bench**, the protocol of
**RelateAnything** ([code](https://github.com/Maelic/RelateAnything) · paper:
*RelateAnything: Real-Time Open-Vocabulary Relation Prediction From Any
Inputs*), packed in the memmap format the evaluators read, plus the
negative-set and cell files the scorers need and the training-side artifacts
the shipped recipe names. **No images are redistributed**: every pack stores
boxes, labels and file names, and the evaluator reads pixels from your copy of
the source dataset (set `RA_DATASETS`).

The protocol has six axes, chosen so that no single one can be won by matching
a benchmark's annotation prior. The definition, the sources, and the reasons
are in
[`benchmark/SPEC.md`](https://github.com/Maelic/RelateAnything/blob/main/benchmark/SPEC.md).

| pack | images | role | axis |
|---|---|---|---|
| `vg150` (test) | 31,128 (26,446 with relations) | in-domain control (100 % vocabulary overlap with VG150-trained baselines) | A1, A3 |
| `psg` (test) | 2,179 | primary transfer, COCO images, 56 predicates | A1, A3, A4, A5 |
| `indoorvg` (test) | 4,403 | domain shift | A1, A3 |
| `hicodet` (test) | 9,490 | 116 external verbs; federated negatives (its `train` split, 33,415 images, is a training source) | A1, A2 |
| `haystack` | 11,368 | explicit negatives on SA-1B images (no image overlap with any training set) | A2 |
| `spatialsense`, `spatialsense_test`, `spatialsense_valid` | 1,920 test images, 2,758 balanced true/false triples | adversarial spatial understanding | A6 |

## Changelog

**2026-11-15: VG150 packs rebuilt from the fixed split.** The `vg150` packs were built from a
release of `maelic/VG150-coco-format` whose `val` split held the first 4,844 images of the test
pool and whose `train` split also held the canonical validation images
([SGG-Benchmark#94](https://github.com/Maelic/SGG-Benchmark/issues/94)). The `test` pack therefore
covered 22,371 of the 26,446 test images with relations, and `val` held test images. All three
packs are now rebuilt from the fixed release (revision `ea6fb3a`): `test` has 31,128 images, of
which 26,446 have relations (the standard VG150 test set), `val` is the standard 5,000-image
validation set, and `train` is the standard training set (57,720 images with relations). Box,
label and relation annotations are unchanged; only the split membership changed. VG150 results
computed with the earlier packs are on a subset of the test set and should be rerun.

## Files

| path | contents |
|---|---|
| `packs/<name>/<split>/` | `meta.json`, `file_names.json`, `img_meta.npy`, `boxes.npy`, `box_cats.npy`, `rels.npy` |
| `datamix/haystack_negatives.json`, `datamix/hicodet_negatives*.json` | explicit negative cells for the A2 precision axis |
| `datamix/spatialsense_{test,valid}_cells.json` | the balanced true/false cells for A6 |
| `datamix/indoorvg_holdout.json` | image ids excluded from training (leakage audit) |
| `datamix/registry.json`, `vg2coco.json`, `psg2coco.json` | id registry used by the converters |
| `detect/yoloworld_ov_psg_test.npz` | the shared YOLO-World detections on PSG test that A4 and A5 score every system on |
| `text_student_v2_512/student.pt` | the distilled predicate text encoder every released checkpoint was trained with |
| `datamix_v22/pair_opportunity.npz`, `datamix_v22/text_space/*` | the soft-supervision and synonym artifacts the training recipe names (`train.sh`) |

## Use

Two commands from a clone of [RelateAnything](https://github.com/Maelic/RelateAnything):

```bash
pip install -e ".[train,hub]"
python benchmark/setup_data.py        # this dataset, the test images (~15 GB), every pack relocated and checked
python benchmark/run_benchmark.py --checkpoint maelic/relsgg-vits16plus
```

`setup_data.py --axes a1 a4 a6` downloads only the images those axes read. The images come
from their sources (VG150, PSG and IndoorVG from `maelic/*-coco-format`, HICO-DET from
`zhimeng/hico_det`, Haystack from its authors, SpatialSense from Zenodo); none are
redistributed here. Each pack's `meta.json` records `img_dir` as `datasets/<source>/<split>`,
which `setup_data.py` re-roots at `$RA_DATASETS`.

`run_benchmark.py` takes a Hugging Face model id or any checkpoint of the recipe and prints
OVS-F1 over A1, A2, A4 and A6. A5, the graph-quality axis, needs a VLM judge (vLLM,
`Qwen/Qwen3.6-27B`) and runs only with `--judge`; it adds the fifth axis of the paper's
composite. The protocol and the composite are defined in
[`benchmark/SPEC.md`](https://github.com/Maelic/RelateAnything/blob/main/benchmark/SPEC.md)
and [`docs/evaluation.md`](https://github.com/Maelic/RelateAnything/blob/main/docs/evaluation.md).

## Licenses

Each pack retains the license of its source: Visual Genome / VG150 and
IndoorVG (CC BY 4.0), PSG (COCO images, CC BY 4.0), HICO-DET (research use),
Haystack (SA-1B research license), SpatialSense (CC BY 4.0). The text student
and the `datamix_v22` artifacts are derivatives of Meta DINOv3 (dino.txt) and
follow the [DINOv3 license](https://ai.meta.com/resources/models-and-libraries/dinov3-license/).

## Citation

```bibtex
@article{neau2026relateanything,
  title   = {RelateAnything: Real-Time Open-Vocabulary Relation Prediction From Any Inputs},
  author  = {Neau, Ma\"elic},
  journal = {arXiv preprint arXiv:2609.12552},
  eprint  = {2609.12552},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CV},
  url     = {https://arxiv.org/abs/2609.12552},
  year    = {2026}
}
```

	---
	license: other
	license_name: mixed-upstream-licenses
	task_categories:
	- object-detection
	language:
	- en
	tags:
	- scene-graph-generation
	- visual-relationship-detection
	- open-vocabulary
	- benchmark
	- evaluation
	pretty_name: OV-SGG-Bench
	size_categories:
	- 10K<n<100K
	---

	# OV-SGG-Bench: evaluation packs for open-vocabulary relation prediction

	The six evaluation sources of OV-SGG-Bench, the protocol of
	RelateAnything ([code](https://github.com/Maelic/RelateAnything) · paper:
	*RelateAnything: Real-Time Open-Vocabulary Relation Prediction From Any
	Inputs*), packed in the memmap format the evaluators read, plus the
	negative-set and cell files the scorers need and the training-side artifacts
	the shipped recipe names. No images are redistributed: every pack stores
	boxes, labels and file names, and the evaluator reads pixels from your copy of
	the source dataset (set `RA_DATASETS`).

	The protocol has six axes, chosen so that no single one can be won by matching
	a benchmark's annotation prior. The definition, the sources, and the reasons
	are in
	[`benchmark/SPEC.md`](https://github.com/Maelic/RelateAnything/blob/main/benchmark/SPEC.md).

	\| pack \| images \| role \| axis \|
	\|---\|---\|---\|---\|
	\| `vg150` (test) \| 31,128 (26,446 with relations) \| in-domain control (100 % vocabulary overlap with VG150-trained baselines) \| A1, A3 \|
	\| `psg` (test) \| 2,179 \| primary transfer, COCO images, 56 predicates \| A1, A3, A4, A5 \|
	\| `indoorvg` (test) \| 4,403 \| domain shift \| A1, A3 \|
	\| `hicodet` (test) \| 9,490 \| 116 external verbs; federated negatives (its `train` split, 33,415 images, is a training source) \| A1, A2 \|
	\| `haystack` \| 11,368 \| explicit negatives on SA-1B images (no image overlap with any training set) \| A2 \|
	\| `spatialsense`, `spatialsense_test`, `spatialsense_valid` \| 1,920 test images, 2,758 balanced true/false triples \| adversarial spatial understanding \| A6 \|

	## Changelog

	2026-11-15: VG150 packs rebuilt from the fixed split. The `vg150` packs were built from a
	release of `maelic/VG150-coco-format` whose `val` split held the first 4,844 images of the test
	pool and whose `train` split also held the canonical validation images
	([SGG-Benchmark#94](https://github.com/Maelic/SGG-Benchmark/issues/94)). The `test` pack therefore
	covered 22,371 of the 26,446 test images with relations, and `val` held test images. All three
	packs are now rebuilt from the fixed release (revision `ea6fb3a`): `test` has 31,128 images, of
	which 26,446 have relations (the standard VG150 test set), `val` is the standard 5,000-image
	validation set, and `train` is the standard training set (57,720 images with relations). Box,
	label and relation annotations are unchanged; only the split membership changed. VG150 results
	computed with the earlier packs are on a subset of the test set and should be rerun.

	## Files

	\| path \| contents \|
	\|---\|---\|
	\| `packs/<name>/<split>/` \| `meta.json`, `file_names.json`, `img_meta.npy`, `boxes.npy`, `box_cats.npy`, `rels.npy` \|
	\| `datamix/haystack_negatives.json`, `datamix/hicodet_negatives*.json` \| explicit negative cells for the A2 precision axis \|
	\| `datamix/spatialsense_{test,valid}_cells.json` \| the balanced true/false cells for A6 \|
	\| `datamix/indoorvg_holdout.json` \| image ids excluded from training (leakage audit) \|
	\| `datamix/registry.json`, `vg2coco.json`, `psg2coco.json` \| id registry used by the converters \|
	\| `detect/yoloworld_ov_psg_test.npz` \| the shared YOLO-World detections on PSG test that A4 and A5 score every system on \|
	\| `text_student_v2_512/student.pt` \| the distilled predicate text encoder every released checkpoint was trained with \|
	\| `datamix_v22/pair_opportunity.npz`, `datamix_v22/text_space/*` \| the soft-supervision and synonym artifacts the training recipe names (`train.sh`) \|

	## Use

	Two commands from a clone of [RelateAnything](https://github.com/Maelic/RelateAnything):

	```bash
	pip install -e ".[train,hub]"
	python benchmark/setup_data.py # this dataset, the test images (~15 GB), every pack relocated and checked
	python benchmark/run_benchmark.py --checkpoint maelic/relsgg-vits16plus
	```

	`setup_data.py --axes a1 a4 a6` downloads only the images those axes read. The images come
	from their sources (VG150, PSG and IndoorVG from `maelic/*-coco-format`, HICO-DET from
	`zhimeng/hico_det`, Haystack from its authors, SpatialSense from Zenodo); none are
	redistributed here. Each pack's `meta.json` records `img_dir` as `datasets/<source>/<split>`,
	which `setup_data.py` re-roots at `$RA_DATASETS`.

	`run_benchmark.py` takes a Hugging Face model id or any checkpoint of the recipe and prints
	OVS-F1 over A1, A2, A4 and A6. A5, the graph-quality axis, needs a VLM judge (vLLM,
	`Qwen/Qwen3.6-27B`) and runs only with `--judge`; it adds the fifth axis of the paper's
	composite. The protocol and the composite are defined in
	[`benchmark/SPEC.md`](https://github.com/Maelic/RelateAnything/blob/main/benchmark/SPEC.md)
	and [`docs/evaluation.md`](https://github.com/Maelic/RelateAnything/blob/main/docs/evaluation.md).

	## Licenses

	Each pack retains the license of its source: Visual Genome / VG150 and
	IndoorVG (CC BY 4.0), PSG (COCO images, CC BY 4.0), HICO-DET (research use),
	Haystack (SA-1B research license), SpatialSense (CC BY 4.0). The text student
	and the `datamix_v22` artifacts are derivatives of Meta DINOv3 (dino.txt) and
	follow the [DINOv3 license](https://ai.meta.com/resources/models-and-libraries/dinov3-license/).

	## Citation

	```bibtex
	@article{neau2026relateanything,
	title = {RelateAnything: Real-Time Open-Vocabulary Relation Prediction From Any Inputs},
	author = {Neau, Ma\"elic},
	journal = {arXiv preprint arXiv:2609.12552},
	eprint = {2609.12552},
	archivePrefix = {arXiv},
	primaryClass = {cs.CV},
	url = {https://arxiv.org/abs/2609.12552},
	year = {2026}
	}
	```