dsaint31 commited on
Commit
b93270f
·
verified ·
1 Parent(s): 4442e50

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +174 -90
README.md CHANGED
@@ -1,103 +1,187 @@
1
  ---
2
- dataset_info:
3
- - config_name: default
4
- features:
5
- - name: row_id
6
- dtype: int32
7
- - name: patient_id
8
- dtype: string
9
- - name: tx_phase
10
- dtype: string
11
- - name: image_r
12
- dtype: image
13
- - name: image_l
14
- dtype: image
15
- - name: class_r
16
- dtype:
17
- class_label:
18
- names:
19
- '0': Angle Class 1
20
- '1': Angle Class 2
21
- '2': Angle Class 3
22
- - name: class_l
23
- dtype:
24
- class_label:
25
- names:
26
- '0': Angle Class 1
27
- '1': Angle Class 2
28
- '2': Angle Class 3
29
- - name: dist_r
30
- dtype: float32
31
- - name: dist_l
32
- dtype: float32
33
- - name: idx
34
- dtype: int32
35
- splits:
36
- - name: train
37
- num_bytes: 33650824
38
- num_examples: 1171
39
- - name: validation
40
- num_bytes: 11461780
41
- num_examples: 390
42
- - name: test
43
- num_bytes: 11570082
44
- num_examples: 390
45
- download_size: 56407499
46
- dataset_size: 56682686
47
- - config_name: single_image
48
- features:
49
- - name: patient_id
50
- dtype: string
51
- - name: tx_phase
52
- dtype: string
53
- - name: image
54
- dtype: image
55
- - name: side
56
- dtype:
57
- class_label:
58
- names:
59
- '0': right
60
- '1': left
61
- - name: labels
62
- dtype:
63
- class_label:
64
- names:
65
- '0': Angle Class 1
66
- '1': Angle Class 2
67
- '2': Angle Class 3
68
- - name: dist
69
- dtype: float32
70
- - name: parent_row_id
71
- dtype: int32
72
- - name: parent_idx
73
- dtype: int32
74
- splits:
75
- - name: train
76
- num_bytes: 34887897
77
- num_examples: 2342
78
- - name: validation
79
- num_bytes: 11452963
80
- num_examples: 780
81
- - name: test
82
- num_bytes: 11237296
83
- num_examples: 780
84
- download_size: 56405697
85
- dataset_size: 57578156
86
  configs:
87
  - config_name: default
 
88
  data_files:
89
  - split: train
90
- path: data/train-*
91
  - split: validation
92
- path: data/validation-*
93
  - split: test
94
- path: data/test-*
95
  - config_name: single_image
96
  data_files:
97
  - split: train
98
- path: single_image/train-*
99
  - split: validation
100
- path: single_image/validation-*
101
  - split: test
102
- path: single_image/test-*
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
103
  ---
 
 
 
 
 
 
 
 
 
 
1
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
  configs:
3
  - config_name: default
4
+ default: true
5
  data_files:
6
  - split: train
7
+ path: "data/train-*"
8
  - split: validation
9
+ path: "data/validation-*"
10
  - split: test
11
+ path: "data/test-*"
12
  - config_name: single_image
13
  data_files:
14
  - split: train
15
+ path: "single_image/train-*"
16
  - split: validation
17
+ path: "single_image/validation-*"
18
  - split: test
19
+ path: "single_image/test-*"
20
+ ---
21
+
22
+ # Malocclusal Occlusal Image Dataset
23
+
24
+ 부정교합 교합 이미지 데이터셋
25
+
26
+ > **Internal Use Only**
27
+ >
28
+ > This dataset repository is private and intended solely for internal research
29
+ > and development. Redistribution or external use is prohibited.
30
+ > 본 데이터셋 저장소는 비공개(private)이며 내부 연구 및 개발 목적으로만 사용됩니다.
31
+ > 외부 배포 또는 재사용은 금지됩니다.
32
+
33
+ ---
34
+
35
+ ## Overview
36
+
37
+ This repository provides occlusal image datasets derived from patient-level records.
38
+ All dataset splits are performed **at the patient level** to prevent information leakage.
39
+ 본 저장소는 환자 단위(patient-level) 기록으로부터 생성된 교합 이미지 데이터셋을 제공합니다.
40
+ 모든 데이터 분할은 **환자 단위**로 수행되어 정보 누출을 방지합니다.
41
+
42
+ The dataset is distributed in **two configurations** within the same Hugging Face
43
+ dataset repository.
44
+ 본 데이터셋은 하나의 Hugging Face dataset repository 내에서
45
+ **두 가지 구성(configuration)** 으로 제공됩니다.
46
+
47
+ Both configurations are **always generated and uploaded together** by the current build script.
48
+ 현재 빌드 스크립트에서는 두 구성 모두가 **항상 함께 생성 및 업로드**됩니다.
49
+
50
+ ---
51
+
52
+ ## Dataset Configurations
53
+ ## 데이터셋 구성
54
+
55
+ ### `default` (pairwise)
56
+
57
+ - One sample contains a **pair of images**: right and left occlusal views.
58
+ 하나의 샘플은 **우측과 좌측 교합 이미지 한 쌍**을 포함합니다.
59
+ - Intended for models that jointly process bilateral occlusal information.
60
+ 양측 교합 정보를 함께 처리하는 모델을 대상으로 합니다.
61
+
62
+ ---
63
+
64
+ ### `single_image`
65
+
66
+ - One sample contains **a single image** (either right or left).
67
+ 하나의 샘플은 **우측 또는 좌측 중 하나의 이미지**만 포함합니다.
68
+ - Generated by exploding the pairwise dataset.
69
+ pairwise 데이터셋을 분해(explode)하여 생성됩니다.
70
+ - Intended for single-view or image-only models.
71
+ 단일 뷰 또는 이미지 단독 입력 모델을 대상으로 합니다.
72
+
73
+ ---
74
+
75
+ ## Splits
76
+
77
+ All configurations provide the same split structure:
78
+ 모든 구성은 동일한 데이터 분할 구조를 가집니다.
79
+
80
+ - `train`
81
+ - `validation`
82
+ - `test`
83
+
84
+ ### Patient-wise split constraint
85
+
86
+ - All samples from the same `patient_id` always belong to the same split.
87
+ 동일한 `patient_id`에 속한 모든 샘플은 항상 동일한 split에 속합니다.
88
+ - This constraint is enforced identically for both configurations.
89
+ 이 제약은 두 구성 모두에 동일하게 적용됩니다.
90
+ - This prevents train–test leakage across configurations.
91
+ 이를 통해 구성 간 train–test 정보 누출을 방지합니다.
92
+
93
+ ---
94
+
95
+ ## Dataset Schema
96
+
97
+ ### Config: `default` (pairwise)
98
+
99
+ | Field | Type | Description |
100
+ |------|------|-------------|
101
+ | `image_r` | `Image` | Right occlusal image / 우측 교합 이미지 |
102
+ | `image_l` | `Image` | Left occlusal image / 좌측 교합 이미지 |
103
+ | `class_r` | `ClassLabel (0..2)` | Angle class (right) / 우측 Angle 분류 |
104
+ | `class_l` | `ClassLabel (0..2)` | Angle class (left) / 좌측 Angle 분류 |
105
+ | `dist_r` | `float32` | Continuous distance (right) / 우측 연속 거리값 |
106
+ | `dist_l` | `float32` | Continuous distance (left) / 좌측 연속 거리값 |
107
+ | `patient_id` | `string` | Patient identifier / 환자 식별자 |
108
+ | `tx_phase` | `string` | Treatment phase / 치료 단계 |
109
+ | `row_id` | `int32` | Pairwise row index / pairwise 행 인덱스 |
110
+ | `idx` | `int32` | Source index (tracking) / 원본 인덱스 |
111
+
112
+ ---
113
+
114
+ ### Config: `single_image`
115
+
116
+ | Field | Type | Description |
117
+ |------|------|-------------|
118
+ | `image` | `Image` | Occlusal image / 교합 이미지 |
119
+ | `side` | `string` | `"r"` or `"l"` / 우측 또는 좌측 |
120
+ | `label` | `ClassLabel (0..2)` | Angle class / Angle 분류 |
121
+ | `dist` | `float32` | Continuous distance / 연속 거리값 |
122
+ | `patient_id` | `string` | Patient identifier / 환자 식별자 |
123
+ | `tx_phase` | `string` | Treatment phase / 치료 단계 |
124
+ | `parent_row_id` | `int32` | Source pairwise row / 원본 pairwise 행 |
125
+ | `parent_idx` | `int32` | Source index (tracking) / 원본 인덱스 |
126
+
127
+ ---
128
+
129
+ ## Label Convention (라벨 규칙)
130
+
131
+ - Angle class labels use **0-based indexing**: `{0, 1, 2}`
132
+ Angle 분류 라벨은 **0부터 시작하는 인덱스** `{0, 1, 2}`를 사용합니다.
133
+ * 0: Angle Class 1
134
+ * 1: Angle Class 2
135
+ * 2: Angle Class 3
136
+ - Labels are converted once during dataset construction.
137
+ 라벨 변환은 데이터셋 생성 시 **한 번만** 수행됩니다.
138
+ - Class semantics follow the standard Angle classification.
139
+ 분류 의미는 표준 Angle 분류 체계를 따릅니다.
140
+
141
+ ---
142
+
143
+ ## Usage (사용 방법)
144
+
145
+ ### Load pairwise dataset (default)
146
+
147
+ ```python
148
+ from datasets import load_dataset
149
+
150
+ ds = load_dataset("dsaint31/hf_maloccu")
151
+ ````
152
+
153
+ ### Load single-image dataset
154
+
155
+ ```python
156
+ from datasets import load_dataset
157
+
158
+ ds_img = load_dataset("dsaint31/hf_maloccu", "single_image")
159
+ ```
160
+
161
+ ---
162
+
163
+ ## Notes (참고 사항)
164
+
165
+ * Dataset splits are **patient-wise**, not sample-wise.
166
+ 데이터 분할은 샘플 단위가 아닌 **환자 단위**로 수행됩니다.
167
+ * Both configurations always share the same patient split.
168
+ 두 구성은 항상 동일한 환자 분할을 공유합니다.
169
+ * Tracking fields (`row_id`, `idx`, `parent_row_id`, `parent_idx`) are provided
170
+ for internal analysis and traceability.
171
+ 추적용 필드(`row_id`, `idx`, `parent_row_id`, `parent_idx`)는
172
+ 내부 분석 및 정합성 검증을 위해 제공됩니다.
173
+ * Images are stored as grayscale PNGs and embedded into parquet files during
174
+ Hub upload (`embed_external_files=True`).
175
+ 이미지는 grayscale PNG로 저장되며 Hub 업로드 시 parquet 파일에
176
+ 직접 포함됩니다(`embed_external_files=True`).
177
+
178
  ---
179
+
180
+ ## License
181
+
182
+ Proprietary – Internal use only.
183
+
184
+ This dataset and any derived artifacts are intended for internal use only.
185
+ Unauthorized copying, distribution, or modification is prohibited.
186
+ 본 데이터셋 및 파생 결과물은 내부 사용만을 목적으로 하며,
187
+ 무단 복제, 배포, 수정은 금지됩니다.