Add files using upload-large-folder tool

README.md

@@ -0,0 +1,217 @@
+---
+license: cc-by-4.0
+task_categories:
+- visual-question-answering
+- image-to-text
+language:
+- en
+tags:
+- multi-view-reasoning
+- robotic-datasets
+- vision-language-model
+- vision-language-action-model
+pretty_name: "XVR: Cross-View Relations"
+size_categories:
+- 100K<n<1M
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: train.jsonl
+  - split: validation
+    path: valid.jsonl
+  - split: test
+    path: xvr_eval.jsonl
+---
+
+# Learning Multi-View Spatial Reasoning from Cross-View Relations
+
+**CVPR 2026**
+
+[Project Page](https://cross-view-relations.github.io) | [arXiv](https://arxiv.org/abs/2603.27967)
+
+XVR (Cross-View Relations) is a large-scale dataset designed to teach Vision-Language Models (VLMs) spatial reasoning across multiple viewpoints. It comprises **100K vision-question-answer samples** derived from **18K diverse 3D scenes** (general domain) and **70K robotic manipulation trajectories** (robotic domain), spanning three fundamental spatial reasoning tasks: **Correspondence**, **Verification**, and **Localization**.
+
+VLMs fine-tuned on XVR achieve clear improvements on multi-view and robotic spatial reasoning benchmarks (MindCube, RoboSpatial), and when used as backbones in Vision-Language-Action (VLA) models, improve manipulation success rates on RoboCasa by ~13% absolute on average.
+
+## Authors
+
+Suchae Jeong\*, Jaehwi Song\*, Haeone Lee, Hanna Kim, Jian Kim, Dongjun Lee, Dong Kyu Shin, Changyeon Kim, Dongyoon Hahm, Woogyeol Jin, Juheon Choi, Kimin Lee
+(KAIST · Config · Hanyang University · Yonsei University · Seoul National University)
+
+\*Equal contribution
+
+## Dataset Structure
+
+```
+XVR/
+├── train.jsonl              # 103,576 training samples
+├── valid.jsonl              #   3,208 validation samples
+├── xvr_eval.jsonl           #   1,866 held-out evaluation samples (XVR-Eval)
+└── image_shards/            # 396,320 images packed into ~40 tar shards
+    ├── images_000.tar       # ~10K images per shard (~2 GB each)
+    ├── images_001.tar
+    └── ...
+```
+
+After extraction, the layout becomes:
+
+```
+XVR/
+├── train.jsonl
+├── valid.jsonl
+├── xvr_eval.jsonl
+└── images/                  # 396,320 image files (extracted from shards)
+    ├── img_0000001.png
+    └── ...
+```
+
+The `prompt_blocks` in each JSONL reference images by the relative path `images/img_XXXXXXX.{png,jpg}`, so the extracted layout matches exactly.
+
+### Splits
+
+| Split | # Samples |
+|---|---|
+| `train` | 103,576 |
+| `validation` (`valid.jsonl`) | 3,208 |
+| `test` (`xvr_eval.jsonl`, XVR-Eval) | 1,866 |
+
+`xvr_eval.jsonl` is the **XVR-Eval benchmark**, constructed from data sources unseen during XVR creation (MobileAloha trajectories and WildRGB-D *boat* category scenes).
+
+### Sample Format
+
+Each line in the `.jsonl` files is a JSON object:
+
+```json
+{
+  "sample_id": "task1_aloha_mobile_episode_000213_sample_1",
+  "task": "task1",
+  "ground_truth_answer": "2",
+  "prompt_blocks": [
+    {"type": "text", "text": "..."},
+    {"type": "image_url", "image_url": {"url": "images/img_0389747.jpg"}},
+    ...
+  ]
+}
+```
+
+- `prompt_blocks`: an ordered list of multimodal blocks (text or image). Image paths are relative to the dataset root (`images/...`).
+- `task`: one of `task1`, `task2`, `task3`, `task4`, `task6`, `task9`, `task10`, `task11` (see task mapping below).
+- `ground_truth_answer`: the expected answer string.
+
+## Task Categories
+
+XVR is organized into three categories and eight specific tasks, inspired by Structure-from-Motion (SfM): identifying correspondences, verifying geometric consistency, and estimating camera poses.
+
+| Category | Task | ID | Description |
+|---|---|---|---|
+| **Correspondence** | Point Correspondence | `task9` | Identify the colored point across views representing the same 3D location |
+| | Directional Correspondence | `task10` | Match directional arrows/vectors consistently across views |
+| **Verification** | Spatial Verification | `task11` | Detect correspondences that violate 3D spatial consistency |
+| | Temporal Verification | `task1` | Identify temporally inconsistent frames in a sequence |
+| **Localization** | Viewpoint Localization | `task6` | Determine which camera view matches a given spatial position |
+| | Directional View Localization | `task2` | Identify the view in a specified direction relative to a reference camera |
+| | Cross-Scenario Localization | `task3` | Match corresponding viewpoints across structurally similar but distinct scenes |
+| | Language-Conditioned Localization | `task4` | Select the view matching a natural-language spatial description |
+
+Each train task contains 12,947 samples; XVR-Eval contains 170–264 samples per task.
+
+## Data Sources
+
+- **General domain** — [WildRGB-D](https://wildrgbd.github.io/): multi-view RGB-D captures with calibrated camera parameters. Only scenes with sufficiently dense point clouds (≥1M points) are retained.
+- **Robotic domain** — [Open X-Embodiment (OXE)](https://robotics-transformer-x.github.io/) and [AgiBot-World](https://agibot-world.com/): we use DROID, MobileAloha, RoboSet, and FMB (the OXE subsets with ≥3 distinct camera views), plus AgiBot-World. Trajectories must be ≥20s with sufficient end-effector motion.
+
+## Usage
+
+### Download
+
+```bash
+# Download the whole dataset (JSONLs + tar shards)
+huggingface-cli download Jaehwisong/XVR --repo-type dataset --local-dir ./XVR
+```
+
+Total size: ~82 GB (image tar shards) + ~150 MB (JSONLs).
+
+### Extracting Image Shards
+
+After downloading, extract all tar shards into a single `images/` folder so that the paths inside `prompt_blocks` (e.g. `images/img_0000001.png`) resolve correctly:
+
+```bash
+cd ./XVR
+mkdir -p images
+for f in image_shards/images_*.tar; do
+  tar -xf "$f" -C images/
+done
+```
+
+Or in parallel (faster):
+
+```bash
+cd ./XVR
+mkdir -p images
+ls image_shards/images_*.tar | xargs -P 8 -I {} tar -xf {} -C images/
+```
+
+Optionally remove the tar shards after extraction to save disk:
+
+```bash
+rm -rf image_shards
+```
+
+### Resolving Image Paths
+
+After extracting the shards, the `prompt_blocks` reference images by relative path (e.g. `images/img_0000001.png`). Prepend the dataset root:
+
+```python
+import json, os
+from PIL import Image
+
+DATA_ROOT = "./XVR"
+with open(os.path.join(DATA_ROOT, "xvr_eval.jsonl")) as f:
+    for line in f:
+        sample = json.loads(line)
+        for block in sample["prompt_blocks"]:
+            if block["type"] == "image_url":
+                img = Image.open(os.path.join(DATA_ROOT, block["image_url"]["url"]))
+                # ... feed to your VLM
+        gt = sample["ground_truth_answer"]
+```
+
+## Results
+
+Qwen3-VL-2B fine-tuned on XVR (**Qwen3-VL-2B-XVR**) achieves a **1.8× relative gain** on XVR-Eval over its base model, ranking first among both open- and closed-source models we evaluated (including GPT-5, Claude-4.5-Sonnet, Gemini-2.5-Pro, Gemini-Robotics-ER-1.5). It also exceeds human performance on Point Correspondence.
+
+| Model | Overall (XVR-Eval) |
+|---|---|
+| Random | 32.64 |
+| Eagle2-2B | 16.99 |
+| PaliGemma2-3B | 17.36 |
+| InternVL-3.5-4B | 32.32 |
+| Qwen3-VL-2B-Instruct | 36.82 |
+| Qwen3-VL-4B-Instruct | 45.02 |
+| Gemini-Robotics-ER-1.5 | 47.48 |
+| Gemini-2.5-Pro | 49.04 |
+| Claude-4.5-Sonnet | 51.18 |
+| Gemini-2.5-Flash | 52.36 |
+| GPT-5 | 61.74 |
+| **Qwen3-VL-2B-XVR (Ours)** | **68.06** |
+| Human | 83.85 |
+
+See the paper for full per-task results and transfer experiments on MindCube-Tiny, RoboSpatial-Home, and RoboCasa VLA manipulation.
+
+## Citation
+
+If you use XVR in your research, please cite:
+
+```bibtex
+@article{jeong2026learning,
+  title={Learning Multi-View Spatial Reasoning from Cross-View Relations},
+  author={Jeong, Suchae and Song, Jaehwi and Lee, Haeone and Kim, Hanna and Kim, Jian and Lee, Dongjun and Shin, Dong Kyu and Kim, Changyeon and Hahm, Dongyoon and Jin, Woogyeol and others},
+  journal={arXiv preprint arXiv:2603.27967},
+  year={2026}
+}
+```
+
+## License
+
+This dataset is released under **CC BY 4.0**. The underlying source datasets (WildRGB-D, OXE subsets, AgiBot-World) retain their respective original licenses; users are responsible for complying with those terms when using derived images.

image_shards/images_000.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eabbe72041c03ef9e1099fea49424e3930b63e3b02e420b67f7f3a6bad4a77b7
+size 1940684800

image_shards/images_001.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b920869c55406dfc807b8f35c983758736d3b9230a33b2d1880bd180803ed092
+size 1947146240

image_shards/images_002.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f71d620b74d5c1af2ca18a402f563d6ce5a936c94691656976890895f65675a
+size 1948200960

image_shards/images_003.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d3db05b2897d2f0997e1d5ed069f40223dee775d5153257ad54f2887ef879f2
+size 1972899840

image_shards/images_004.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2e59c59848fe5fc1b666a74cd150d1ccb234cea36b0dbf11cb1fdb85e36beeb
+size 2009282560

image_shards/images_005.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e74179102579cb57f090830ae5ffb7157b9f51191a39e3204ef80314beea94af
+size 2004039680

image_shards/images_006.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd79c3d259b141352f5d9144dd55e57c7700c1a771e535b864fe522a29308480
+size 1959710720

image_shards/images_007.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f4be0846720723ab4e92f64d04529a807a21a9667504a8706f217250360e347
+size 2021744640

image_shards/images_008.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71c069910e5f4e7ecd32478d0c0e15f45b0800feb3bb8e0cf0ed858f8e85585a
+size 2039490560

image_shards/images_009.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e8de693e1e0459d6b82d43a43453cf8e4ff2128dfb02c58b32b046a3fe4d469
+size 2030776320

image_shards/images_010.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5cba0ff04dbab40dbdef675d32a2cee53b7190f82f660a6bee6aa61556a51888
+size 2108262400

image_shards/images_011.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35eaf718e33f8f5c014b3138c4f1a7ca977ddef54269c3f5701a15a180ebf344
+size 2141173760

image_shards/images_012.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0069803bd3eba0ec324be0de1b598d0b1f9e726e4fc77c32ee1a7ffe278da9d5
+size 2056990720

image_shards/images_013.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28313d71235ee9ba348d11e2a4f4f9cc2de36899f83b85fdbb9b13dd71883da7
+size 2132008960

image_shards/images_014.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbb623922c24fa2dde32b7cce688f9ab549cbed3ae5af5cf0630aaedb2895f93
+size 2050498560

image_shards/images_015.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:669663ef884db9a0bb768a5f43acdc510d8f9e16fda8093ecc14a03521bde919
+size 2187294720

image_shards/images_016.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:332435e04e9753406e9f3c5e83221e71bb82355923b09fdd7bea4c1a32c90717
+size 2158264320

image_shards/images_017.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed43de6b0438e2b1dbaa3e89fc937541c372f3c8c20058de41a4ae866e359a89
+size 2150512640

image_shards/images_018.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ce383bdce972bd73229b201bcbc242b2eb6e549ca6d451f81da0210e87f0cb8
+size 2182676480

image_shards/images_019.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14e6ea8fc47b50c69093892f407edc378536d354e59606536961246d7506d48e
+size 2184683520

image_shards/images_020.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6eb20b304cb80ec9470a8032470eb8a8e0b27554909f812bb973ecf771243f11
+size 2170992640

image_shards/images_021.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b15fdfd28c7a5d6d00dc4388c4576602a9f15c33be3c43581b6267913ced9070
+size 2243829760

image_shards/images_022.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75ce88c764babd56674db06a6fa21c72e2601fbe1ea3db2812eb61c05325a473
+size 2215628800

image_shards/images_023.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:034851546e5db06e18c95c20fa79edfef694e944ee5bf9fa0d473418e175e37d
+size 2285076480

image_shards/images_024.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f766069eb469f0bcf46696641fa8c5a54ca31209d05f1f7d822b710c17f7b18
+size 2217922560

image_shards/images_025.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7577bc11f8569f9a16e6e1f866d2b0a2e7297142843fee4b499deb960905b771
+size 2279403520

image_shards/images_026.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54ed9e512fb260bea00268d743ec8caf1139109f536ff26d4b8bf1a88152bb0c
+size 2246942720

image_shards/images_027.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a05b918808245cd5f5841878dcd966906218d2e313547170ebd471354c87b95
+size 2331084800

image_shards/images_028.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd37bfe3077aa4321e400912cda69d8c885900601902b885f8920f09b5710c20
+size 2333849600

image_shards/images_029.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c920a3d0ab1e67199df738f0e76d79b2f4b7fafe0b56d9f93d3c6d5ecc5e4d79
+size 2294528000

image_shards/images_030.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e52b8435a1b36463d2db8db77989123aa351593bf1c59bc36adf029784eeeb70
+size 2326179840

image_shards/images_031.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5efd469cc3a64b37a0a46a205d906a21029f0b8e33b5a41acfb5fd85fa0690e0
+size 2316339200

image_shards/images_032.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b96cb353a8de7718068b68187e70ae4351b57fd13ee5ef929993346a263ac2de
+size 2367109120

image_shards/images_033.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bfb64ed228bbcdb704ef9a7087fa49fefbe8b9e8eb3b1364fe9b98f5b0bff43
+size 2385704960

image_shards/images_034.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:646fb790d3924a805d717caa6c1fe1b40406aeeac7dc65faca208934c736b56e
+size 2336532480

image_shards/images_035.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f45df74908617eca27eff6c1770da3d25a65d9211676ad9268c8f4ba91d6a30
+size 2394583040

image_shards/images_036.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6280be64f46fc7c4dd9354bc321613bdd23b02d392c12bda1c0c029e232aeead
+size 2431057920

image_shards/images_037.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4aae4005b83ac8ebfc8ef5f3b96f94a613f445e93d50a878a6ded0d41d746c07
+size 2440110080

image_shards/images_038.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9363867ac4d7f1e2b0e12720bd16bff4f32cc22159b97510c66eb853c0a761da
+size 2351656960

image_shards/images_039.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b22e73e236d37f6c8f8ee8a4cdc81a0134d27f57a9c11d44f49fc875980d4642
+size 1674874880