suijinru commited on
Commit
a6e9bef
·
verified ·
1 Parent(s): 823734e

Update dataset card and task previews

Browse files
Files changed (2) hide show
  1. README.md +33 -5
  2. SHA256SUMS +1 -1
README.md CHANGED
@@ -1,5 +1,5 @@
1
  ---
2
- pretty_name: MultiView-Bench Synthetic
3
  task_categories:
4
  - visual-question-answering
5
  language:
@@ -16,12 +16,15 @@ configs:
16
  drop_labels: true
17
  ---
18
 
19
- # MultiView-Bench Synthetic
20
 
21
- Synthetic image-and-text portion of **MultiView-Bench: A Diagnostic Benchmark
22
- for World-Centric Multi-View Integration in VLMs**
23
  ([arXiv:2607.08970](https://arxiv.org/abs/2607.08970)).
24
 
 
 
 
25
  ## Contents
26
 
27
  - 2,400 evaluation samples across 24 benchmark variants
@@ -35,6 +38,31 @@ It uses 3DCoMPaT++ source assets and is not part of this release. This package
35
  also contains no GLTF, BIN, model asset, source code, credentials, or model
36
  outputs.
37
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
38
  ## Load
39
 
40
  After upload:
@@ -42,7 +70,7 @@ After upload:
42
  ```python
43
  from datasets import load_dataset
44
 
45
- dataset = load_dataset("suijinru/MultiViewBench-Synthetic", split="test")
46
  sample = dataset[0]
47
  print(sample["benchmark"], sample["question"], sample["answer"])
48
  print(sample["images"])
 
1
  ---
2
+ pretty_name: MultiView-Bench
3
  task_categories:
4
  - visual-question-answering
5
  language:
 
16
  drop_labels: true
17
  ---
18
 
19
+ # MultiView-Bench
20
 
21
+ Evaluation data for **MultiView-Bench: A Diagnostic Benchmark for
22
+ World-Centric Multi-View Integration in VLMs**
23
  ([arXiv:2607.08970](https://arxiv.org/abs/2607.08970)).
24
 
25
+ This repository currently contains the synthetic image-and-text portion of the
26
+ benchmark. The real-world subset is not included; see the scope note below.
27
+
28
  ## Contents
29
 
30
  - 2,400 evaluation samples across 24 benchmark variants
 
38
  also contains no GLTF, BIN, model asset, source code, credentials, or model
39
  outputs.
40
 
41
+ ## Task previews
42
+
43
+ Each sample asks for a relative 3D position and must be solved by integrating
44
+ the evidence across its views.
45
+
46
+ ### Colored-object three-view task
47
+
48
+ **Question:** What is the relative position of the olive cone to the indigo
49
+ cube?
50
+ **Answer:** `(-X, -Y, -Z)`
51
+
52
+ | Front | Side | Top |
53
+ |:---:|:---:|:---:|
54
+ | ![Front view of colored-object task 17](test/images/exp2D_three_views_2d_colors/task_17/task_17_front.png) | ![Side view of colored-object task 17](test/images/exp2D_three_views_2d_colors/task_17/task_17_side.png) | ![Top view of colored-object task 17](test/images/exp2D_three_views_2d_colors/task_17/task_17_top.png) |
55
+
56
+ ### 3D isometric task
57
+
58
+ **Question:** What is the relative position of the cylinder to the cone?
59
+ **Answer:** `(0X, 0Y, -Z)`
60
+ The full sample contains six views; three are shown here.
61
+
62
+ | View 0 | View 2 | View 4 |
63
+ |:---:|:---:|:---:|
64
+ | ![Isometric view 0 of 3D task 7](test/images/exp3D_3d_isometric_two_axis_fixed/task_7/task_7_3d_view_00.png) | ![Isometric view 2 of 3D task 7](test/images/exp3D_3d_isometric_two_axis_fixed/task_7/task_7_3d_view_02.png) | ![Isometric view 4 of 3D task 7](test/images/exp3D_3d_isometric_two_axis_fixed/task_7/task_7_3d_view_04.png) |
65
+
66
  ## Load
67
 
68
  After upload:
 
70
  ```python
71
  from datasets import load_dataset
72
 
73
+ dataset = load_dataset("suijinru/MultiViewBench", split="test")
74
  sample = dataset[0]
75
  print(sample["benchmark"], sample["question"], sample["answer"])
76
  print(sample["images"])
SHA256SUMS CHANGED
@@ -1,5 +1,5 @@
1
  2d178ef94980ddaf2629c3919a4cae207fa354adc42a23d23096a1ed6c60dc67 MANIFEST.json
2
- 541318143cb7db31ef1c99cfcde47f408f8202c1edf67b9bffe65fb515d9adf9 README.md
3
  5db318623cac9919e2496c5dba9be2acf87d589f09397960e10be24453263abb test/images/exp2D_three_views_2d/task_1/task_1_front.png
4
  366d83cf773f388583f7c5a334e06f820fe01b3694b227f3d6fdb8d0f940945e test/images/exp2D_three_views_2d/task_1/task_1_side.png
5
  d02da112f8bcc0abfe83845c2a092f2e8ed990dc7aa143f58bf0dec4303f081d test/images/exp2D_three_views_2d/task_1/task_1_top.png
 
1
  2d178ef94980ddaf2629c3919a4cae207fa354adc42a23d23096a1ed6c60dc67 MANIFEST.json
2
+ 379369389d3e27880c2a832031030470b04c7ab9d6c44cb88cdd3ef8ab1e4d96 README.md
3
  5db318623cac9919e2496c5dba9be2acf87d589f09397960e10be24453263abb test/images/exp2D_three_views_2d/task_1/task_1_front.png
4
  366d83cf773f388583f7c5a334e06f820fe01b3694b227f3d6fdb8d0f940945e test/images/exp2D_three_views_2d/task_1/task_1_side.png
5
  d02da112f8bcc0abfe83845c2a092f2e8ed990dc7aa143f58bf0dec4303f081d test/images/exp2D_three_views_2d/task_1/task_1_top.png