Datasets
The dataset list and creation (direct upload, external source), the Overview and Files tabs of the detail, and file previews (image, PLY 3D, time series)
Required permission: View
This is where you keep the data used for training. A dataset is usable as soon as it is created, and afterwards you can freely add or delete files. When you upload files, the server validates them one by one to determine their kind (image, annotation, point cloud, time series, tabular), pairs images with annotations, and automatically rewrites the dataset summary (manifest).
Every member can view; creating, uploading files and deleting need dataset_write (DEVELOPER, OPERATOR, ADMIN).
Dataset list
| # | Area | Description |
|---|---|---|
| ① | Search | Find by name. |
| ② | Validation filter | All · Passed · Partial errors · Failed (전체 · 통과 · 일부 오류 · 실패). |
| ③ | External sources (외부 소스) | Opens the list of registered external data servers (DataOps). |
| ④ | New dataset (새 데이터셋) | Creation dialog (dataset_write). |
| ⑤ | Validation | Badge with the validation result of the dataset's files. |
In the table, Modality (모달리티) is the kind of data (point cloud, image, time series, tabular, mixed), and Source (소스) is a note about where the data came from. Datasets linked to an external source get a Sync (동기화) badge in the source column.
New dataset
Choose the method with the two cards at the top of the dialog.
Direct upload (직접 업로드) — set only the name, modality and source (optional) to create an empty dataset, then upload files on the detail screen. Time series and tabular data are both CSV/Parquet, so the modality you choose here decides the kind (time series is a measurement history with a time axis; tabular is one row per subject).
Sync from an external source (외부 소스에서 동기화) — choose a remote dataset on a registered external server to create the dataset, and the first sync starts right away.
The external source list (the External sources (외부 소스) button) shows the registered servers and the number of linked datasets, and lets you check the connection, edit and delete. The procedure for registering an external source and syncing is described in the Integration guide.
Dataset detail — Overview
| # | Area | Description |
|---|---|---|
| ① | Summary | Name, validation badge, modality, number of files, size, and Trainable (학습 가능) (number of files training will use) or Not trainable (학습 불가). |
| ② | Actions | Train with this dataset (이 데이터셋으로 학습) (develop — opens the wizard with this dataset), Edit (편집) (name, source, modality), Revalidate (재검증) (checks all files again), Delete (삭제). Edit, revalidate and delete need dataset_write. |
| ③ | Tabs | Overview (개요) · Files (파일). |
| ④ | External sources (외부 소스) | For a dataset managed by upload, Connect external source (외부 소스 연결); for a linked dataset, the last sync and a Sync (동기화) button. |
| ⑤ | File summary (파일 집계) | Total, valid, error, total size, number of files per kind, number of images without a label pair. |
| ⑥ | Manifest (매니페스트) | Dataset summary such as source, record count and storage location. |
Further down come the validation results (검증 결과) (schema conformance, label match rate and so on) and the training usage history (학습 사용 이력) (runs and model versions that used this dataset).
Dataset detail — Files
| # | Area | Description |
|---|---|---|
| ① | Upload | Drag and drop files or folders, or click to choose. Supported formats depend on the modality; to upload many at once, bundle them in a ZIP. Folders must be bundled in a ZIP. |
| ② | Filters | Kind (image, annotation, point cloud…) and status (valid, error, validating, uploaded). Datasets linked to an external source have an extra origin filter (upload, sync). |
| ③ | Sort and view | Newest, oldest, name, largest; grid/table toggle. |
| ④ | File card | Thumbnail, name, status, size, kind, origin. Images get Labeled (라벨 있음) / Unlabeled (라벨 없음) (whether a paired annotation exists). Click to open the file preview. |
In the table view, select several files and delete them at once with Delete selected (선택 삭제). The Summary (요약) column shows a per-file summary such as size, labels and match rate.
File preview
The preview differs by file kind. Use the previous and next arrows at the top to move through the list, and Download (다운로드) and Delete (삭제).
| Kind | Preview |
|---|---|
| Image | The original image. If a paired annotation (LabelMe, COCO, VOC) exists, boxes, polygons and labels are drawn over it; turn them off with Hide annotations (주석 숨김). |
| Point cloud (PLY) | 3D viewer — shows the preview points the server reduced, colored by label. |
| Time series | Numeric columns (up to 6) as line charts along the time axis. |
| Tabular | The first rows as a table. |
| JSON annotation | Indented JSON (2 MB or less). |
| Other and error files | "No preview (미리보기 없음)" and the validation error message. |
| # | Area | Description |
|---|---|---|
| ① | Navigation and actions | Previous and next file (position/total), Download (다운로드), Delete (삭제). |
| ② | Annotation toggle | Hides or shows the overlaid annotations again. The number in parentheses is the shape count. |
| ③ | Metadata (메타데이터) | Kind, size, Content-Type, summary, SHA-256, upload time, label pair info and Recompute matching (매칭 다시 계산). |
PLY 3D viewer:
| # | Area | Description |
|---|---|---|
| ① | 3D canvas | Drag to rotate, wheel to zoom, right-drag to pan. At the top is a legend of colors and point counts per label. |
| ② | Controls | The viewer is a downsampled preview and the label colors are fixed. Get the original with Download (다운로드). |
Time-series CSV:
Statuses and badges
Dataset validation badge:
| Badge | Value | Meaning |
|---|---|---|
| Passed (통과) | PASS | All files passed validation |
| Partial errors (일부 오류) | PARTIAL | One or more files have errors — hover to see how many |
| Failed (실패) | FAIL | Appears only when an external system passes a failure verdict through the manifest |
| No files (파일 없음) | — | No files yet |
File status:
| Badge | Value | Meaning |
|---|---|---|
| Uploaded (업로드됨) | UPLOADED | Stored and waiting for validation |
| Validating (검증 중) | VALIDATING | The server is checking it (the list refreshes automatically) |
| Valid (유효) | VALID | Passed — used for training and previews |
| Error (오류) | INVALID | Validation failed — the error message appears in red in the table view |
External sync statuses are Pending · Syncing · Completed · Failed · Canceled (대기 · 동기화 중 · 완료 · 실패 · 취소됨).