Susan0803 commited on
Commit
f0e2000
·
verified ·
1 Parent(s): 765eb0f

Add anonymous visual perception evaluation dataset

Browse files
README.md ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ pretty_name: Anonymous Visual Perception Benchmark
3
+ task_categories:
4
+ - image-text-to-text
5
+ language:
6
+ - zh
7
+ - en
8
+ size_categories:
9
+ - 1K<n<10K
10
+ configs:
11
+ - config_name: default
12
+ data_files:
13
+ - split: test
14
+ path: data/test-*.parquet
15
+ ---
16
+
17
+ # Anonymous Visual Perception Benchmark
18
+
19
+ This evaluation dataset contains 2,876 image-question pairs with Chinese and
20
+ English questions, reference answers, and hierarchical visual-perception labels.
21
+ Images are embedded in Parquet files; no external image service is required.
22
+
23
+ ## Data Format
24
+
25
+ The dataset has one `test` split. Each row contains:
26
+
27
+ | Field | Description |
28
+ | --- | --- |
29
+ | `id` | Release-local sample identifier, unrelated to internal identifiers. |
30
+ | `image` | Embedded PNG image, decoded as a Hugging Face `Image` feature. |
31
+ | `question_zh`, `question_en` | Chinese and English question text. |
32
+ | `answer_zh`, `answer_en` | Chinese and English reference answers. |
33
+ | `answer_mask` | JSON-encoded geometric annotations where provided; otherwise an empty string. |
34
+ | `l1_zh` through `l4_zh` | Chinese hierarchical category labels. |
35
+ | `l1_en` through `l4_en` | Corresponding English category labels. |
36
+
37
+ The seven top-level categories are Grounding, Semantic Recognition,
38
+ Fine-grained Discrimination, Visual Tracking, Space Understanding, OCR, and
39
+ Structured Visual Perception. There are 21 distinct top-level/subcategory pairs.
40
+ The OCR portion includes 300 questions labeled Text Recognition in Less Common
41
+ Languages.
42
+
43
+ The original annotation text is preserved. Some lower-level labels are absent:
44
+ 7 rows have an empty third-level label and 75 rows have an empty fourth-level
45
+ label. Empty annotations are not inferred or fabricated.
46
+
47
+ ## Loading
48
+
49
+ ```python
50
+ from datasets import load_dataset
51
+
52
+ dataset = load_dataset("Susan0803/visual-perception-benchmark-anon", split="test")
53
+ sample = dataset[0]
54
+ image = sample["image"]
55
+ question = sample["question_en"]
56
+ reference_answer = sample["answer_en"]
57
+ ```
58
+
59
+ To load a local checkout, use:
60
+
61
+ ```python
62
+ dataset = load_dataset("parquet", data_files={"test": "data/test-*.parquet"}, split="test")
63
+ ```
64
+
65
+ ## Release Notes
66
+
67
+ Only questions, reference annotations, images, and category labels are included.
68
+ Model predictions, model identifiers, evaluation scores, reviewer notes,
69
+ internal storage addresses, and internal sample identifiers are excluded.
70
+
71
+ Images are stored as self-contained PNGs without embedded EXIF or text metadata.
72
+ Image orientation is applied to pixels, and embedded color profiles are converted
73
+ to sRGB. Multi-picture JPEG containers are represented by their primary image.
74
+ This removes file metadata, not text visibly present within benchmark images.
75
+ The images themselves are evaluation content and are not redacted.
76
+
77
+ This repository does not assert an additional license over third-party image
78
+ content. Public accessibility alone does not establish redistribution rights.
data/test-00000-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95f8424786390ad160e81477afb31bd21c59f3a20cb7a81bff7fda339b27e69e
3
+ size 106152732
data/test-00001-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43cf0e46dd114911d8db7f8a896911c2dbc62575b1f92d8bf5fc11ed466d53a1
3
+ size 174506281
data/test-00002-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:213449e805963b274894cabd76b065e2aa9f5535838315c934b7351e4c10525b
3
+ size 96469655
data/test-00003-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ff6f7efe3b2a05c8e1f45446ee691ede6104460583f3b4a765a0bd22ae813742
3
+ size 41569141
data/test-00004-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad6f64e1ff38280c400df49be35bcde6ccb915e773e1477303730f8850954cb5
3
+ size 51230572
data/test-00005-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5989175f7f41ef9682e9df2831a8a7d438efe6be6d3bc1a63aaade3bbbe7284c
3
+ size 114072538
data/test-00006-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df2f612ee0532a743ffcbb5ae2306981685f9326bd162789c4f0c1b959da6875
3
+ size 43418289
data/test-00007-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ece8e466bd3069f85e4a0cb9c3520b185fc30915ef57b1c8836327e7c36fca9b
3
+ size 61654378
data/test-00008-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0afbce9a94c1e5d483a80066dcd064bad2864d51e072b5a09be2c2ed057c4325
3
+ size 35635387
data/test-00009-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:074db264b50c5efcccd6d3d1d924c50a6ef34d746c940096b19e04b2e06cd933
3
+ size 67452013
data/test-00010-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1fd0115206e10072423748773d4babc812d8420418182ea881a1d964dc63c204
3
+ size 51855505
data/test-00011-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:524b9af1936000904ed0482703961d9feda60b9dae6948ccc0dca3c89d6ca631
3
+ size 92341554
data/test-00012-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc5fc2ccacb3b75efbe94ff3482638f337a11663e06c7f8291a2f7c8bb3ceda6
3
+ size 47431239
data/test-00013-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2780a13048b90d7b495be437f2d2ca9c65f4bffa574877baba69b3d61ab578bc
3
+ size 84950014
data/test-00014-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3e576212a23219638d6e045d086c41a61df5f261e5099261a581d1d9bbc0f04
3
+ size 102738606
data/test-00015-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:048cd903a7c9448b7463969b5184abf4f50c8f44922efedcf551958ada8fa17b
3
+ size 16453897
data/test-00016-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc137686bf87e71a2124f375955781ab4a22c54af4aba0a3ae5e68490168e7d0
3
+ size 13564791
data/test-00017-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:455b9642e8d3a0ebef704c78c83a20aed39cf12e1e6ad214d3b4bef08109aa41
3
+ size 30679779
data/test-00018-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f20cebfdf39ac27d1da3f3a90cfa4eaec9b39c3607b105e3c6278795349a4e5
3
+ size 89206111
data/test-00019-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9672124490154b5a39850fc10eec1eba85a3ba17cc0581ee33b1275282a5dad
3
+ size 221528192
data/test-00020-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:711bfd02951b297fd77a86cecb335c782d8b9f3912fca71819f68652b4cedb0b
3
+ size 219234833
data/test-00021-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b4a7802f4e4b5a6c4723c7044f0d6a124b90678924c0752bacf08d948a192f3
3
+ size 160738757
data/test-00022-of-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a493e815541afc4b0e55edf5d9c9410f9e6f3588c40fa586f2e7c1c8cb008bb8
3
+ size 59301823