GeorgyG commited on
Commit
85592a6
·
verified ·
1 Parent(s): e204166

Update README with new statistics (removed screenshots)

Browse files
Files changed (1) hide show
  1. README.md +13 -18
README.md CHANGED
@@ -51,8 +51,8 @@ This is the validation split - publicly available for open evaluation and compar
51
 
52
  ## 📊 Dataset Statistics
53
 
54
- - **Total samples:** 1,610
55
- - **Unique images:** 550
56
  - **Task types:** 5
57
 
58
  ---
@@ -61,22 +61,17 @@ This is the validation split - publicly available for open evaluation and compar
61
 
62
  ```
63
  MWS-Vision-Bench/
64
- ├── annotations.json # Dataset annotations
65
  ├── images/ # Image files organized by category
66
  │ ├── business/
67
  │ │ ├── scans/
68
  │ │ ├── sheets/
69
  │ │ ├── plans/
70
  │ │ └── diagramms/
71
- ── personal/
72
- ├── hand_documents/
73
- ├── hand_notebooks/
74
- └── hand_misc/
75
- │ └── screenshots/
76
- │ ├── business_apps/
77
- │ ├── cad_code/
78
- │ ├── games/
79
- │ └── mts_apps/
80
  └── README.md # This file
81
  ```
82
 
@@ -103,11 +98,11 @@ Each entry in `annotations.json` follows this structure:
103
 
104
  | Task | Description | Count |
105
  |------|--------------|-------|
106
- | `document parsing ru` | Parsing structured documents | 247 |
107
- | `full-page OCR ru` | End-to-end OCR on full pages | 154 |
108
- | `key information extraction ru` | Extracting key fields | 121 |
109
- | `reasoning VQA ru` | Visual reasoning in Russian | 550 |
110
- | `text grounding ru` | Text–region alignment | 538 |
111
 
112
  ---
113
 
@@ -166,7 +161,7 @@ If you use this dataset in your research, please cite:
166
  ## 🇷🇺 Краткое описание
167
 
168
  **MWS Vision Bench** — первый русскоязычный бенчмарк для бизнес-OCR в эпоху мультимодальных моделей.
169
- Он включает 1610 примеров и 5 типов задач, отражающих реальные сценарии обработки документов, интерфейсов и рукописных данных.
170
  Датасет создан для оценки и развития мультимодальных LLM в русскоязычном контексте.
171
  📄 *Научная статья в процессе подготовки (paper coming soon).*
172
 
 
51
 
52
  ## 📊 Dataset Statistics
53
 
54
+ - **Total samples:** 1,302
55
+ - **Unique images:** 400
56
  - **Task types:** 5
57
 
58
  ---
 
61
 
62
  ```
63
  MWS-Vision-Bench/
64
+ ├── metadata.jsonl # Dataset annotations
65
  ├── images/ # Image files organized by category
66
  │ ├── business/
67
  │ │ ├── scans/
68
  │ │ ├── sheets/
69
  │ │ ├── plans/
70
  │ │ └── diagramms/
71
+ ── personal/
72
+ ├── hand_documents/
73
+ ├── hand_notebooks/
74
+ └── hand_misc/
 
 
 
 
 
75
  └── README.md # This file
76
  ```
77
 
 
98
 
99
  | Task | Description | Count |
100
  |------|--------------|-------|
101
+ | `document parsing ru` | Parsing structured documents | 243 |
102
+ | `full-page OCR ru` | End-to-end OCR on full pages | 144 |
103
+ | `key information extraction ru` | Extracting key fields | 119 |
104
+ | `reasoning VQA ru` | Visual reasoning in Russian | 400 |
105
+ | `text grounding ru` | Text–region alignment | 396 |
106
 
107
  ---
108
 
 
161
  ## 🇷🇺 Краткое описание
162
 
163
  **MWS Vision Bench** — первый русскоязычный бенчмарк для бизнес-OCR в эпоху мультимодальных моделей.
164
+ Он включает 1302 примера и 5 типов задач, отражающих реальные сценарии обработки бизнес-документов и рукописных данных.
165
  Датасет создан для оценки и развития мультимодальных LLM в русскоязычном контексте.
166
  📄 *Научная статья в процессе подготовки (paper coming soon).*
167