q792602257/DataMate
0
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

feat(annotation): 支持音频和视频数据类型的标注任务

Browse Source 
- 添加了音频和视频数据类型常量定义
- 实现了音频和视频标注模板的内置配置
- 扩展前端组件以支持按数据类型过滤标注模板
- 重构后端编辑器服务以处理音频和视频任务构建
- 更新数据库初始化脚本包含音频和视频标注模板
- 添加音频和视频数据类型的预览URL映射逻辑
This commit is contained in:
Jerry Yan
2026-01-26 23:54:40 +08:00
parent 47295e8cdf
commit 977a930c97
5 changed files with 461 additions and 59 deletions
Download Patch File Download Diff File Expand all files Collapse all files

97
runtime/datamate-python/app/module/annotation/service/editor.py
View File

@@ -43,6 +43,8 @@ logger = get_logger(__name__)
TEXT_DATA_KEY = "text"
IMAGE_DATA_KEY = "image"
AUDIO_DATA_KEY = "audio"
VIDEO_DATA_KEY = "video"
DATASET_ID_KEY = "dataset_id"
FILE_ID_KEY = "file_id"
FILE_NAME_KEY = "file_name"
@@ -53,9 +55,19 @@ SEGMENT_INDEX_KEY = "segment_index"
SEGMENT_INDEX_CAMEL_KEY = "segmentIndex"
JSONL_EXTENSION = ".jsonl"
TEXTUAL_OBJECT_CATEGORIES = {"text", "document"}
MEDIA_OBJECT_CATEGORIES = {"image"}
IMAGE_OBJECT_CATEGORIES = {"image"}
MEDIA_OBJECT_CATEGORIES = {"media"}
OBJECT_NAME_HEADER_PREFIX = "dm_object_header_"
SUPPORTED_EDITOR_DATASET_TYPES = ("TEXT", "IMAGE")
DATASET_TYPE_TEXT = "TEXT"
DATASET_TYPE_IMAGE = "IMAGE"
DATASET_TYPE_AUDIO = "AUDIO"
DATASET_TYPE_VIDEO = "VIDEO"
SUPPORTED_EDITOR_DATASET_TYPES = (
DATASET_TYPE_TEXT,
DATASET_TYPE_IMAGE,
DATASET_TYPE_AUDIO,
DATASET_TYPE_VIDEO,
)
SEGMENTATION_ENABLED_KEY = "segmentation_enabled"
@@ -174,21 +186,19 @@ class AnnotationEditorService:
return keys[0]
@classmethod
def _resolve_primary_media_key(
def _resolve_media_value_keys(
cls,
label_config: Optional[str],
default_key: str,
categories: Optional[set[str]] = None,
) -> str:
) -> List[str]:
if not label_config:
return default_key
return [default_key]
target_categories = categories or set()
keys = cls._extract_object_value_keys_by_category(label_config, target_categories)
if not keys:
return default_key
if default_key in keys:
return default_key
return keys[0]
return [default_key]
return keys
@staticmethod
def _try_parse_json_payload(text_content: str) -> Optional[Dict[str, Any]]:
@@ -467,7 +477,10 @@ class AnnotationEditorService:
dataset_type = self._normalize_dataset_type(await self._get_dataset_type(project.dataset_id))
if dataset_type not in SUPPORTED_EDITOR_DATASET_TYPES:
raise HTTPException(status_code=400, detail="当前仅支持 TEXT/IMAGE 项目的内嵌编辑器")
raise HTTPException(
status_code=400,
detail="当前仅支持 TEXT/IMAGE/AUDIO/VIDEO 项目的内嵌编辑器",
)
file_result = await self.db.execute(
select(DatasetFiles).where(
@@ -479,9 +492,15 @@ class AnnotationEditorService:
if not file_record:
raise HTTPException(status_code=404, detail=f"文件不存在或不属于该项目: {file_id}")
if dataset_type == "IMAGE":
if dataset_type == DATASET_TYPE_IMAGE:
return await self._build_image_task(project, file_record, file_id)
if dataset_type == DATASET_TYPE_AUDIO:
return await self._build_audio_task(project, file_record, file_id)
if dataset_type == DATASET_TYPE_VIDEO:
return await self._build_video_task(project, file_record, file_id)
return await self._build_text_task(project, file_record, file_id, segment_index)
async def _build_text_task(
@@ -668,23 +687,20 @@ class AnnotationEditorService:
currentSegmentIndex=current_segment_index,
)
async def _build_image_task(
async def _build_media_task(
self,
project: LabelingProject,
file_record: DatasetFiles,
file_id: str,
default_key: str,
categories: set[str],
) -> EditorTaskResponse:
label_config = await self._resolve_project_label_config(project)
image_key = self._resolve_primary_media_key(
label_config,
IMAGE_DATA_KEY,
MEDIA_OBJECT_CATEGORIES,
)
media_keys = self._resolve_media_value_keys(label_config, default_key, categories)
preview_url = self._build_file_preview_url(project.dataset_id, file_id)
file_name = str(getattr(file_record, "file_name", ""))
task_data: Dict[str, Any] = {
image_key: preview_url,
FILE_ID_KEY: file_id,
FILE_ID_CAMEL_KEY: file_id,
DATASET_ID_KEY: project.dataset_id,
@@ -692,6 +708,9 @@ class AnnotationEditorService:
FILE_NAME_KEY: file_name,
FILE_NAME_CAMEL_KEY: file_name,
}
for key in media_keys:
task_data[key] = preview_url
self._apply_text_placeholders(task_data, label_config)
# 获取现有标注
ann_result = await self.db.execute(
@@ -738,6 +757,48 @@ class AnnotationEditorService:
currentSegmentIndex=0,
)
async def _build_image_task(
self,
project: LabelingProject,
file_record: DatasetFiles,
file_id: str,
) -> EditorTaskResponse:
return await self._build_media_task(
project=project,
file_record=file_record,
file_id=file_id,
default_key=IMAGE_DATA_KEY,
categories=IMAGE_OBJECT_CATEGORIES,
)
async def _build_audio_task(
self,
project: LabelingProject,
file_record: DatasetFiles,
file_id: str,
) -> EditorTaskResponse:
return await self._build_media_task(
project=project,
file_record=file_record,
file_id=file_id,
default_key=AUDIO_DATA_KEY,
categories=MEDIA_OBJECT_CATEGORIES,
)
async def _build_video_task(
self,
project: LabelingProject,
file_record: DatasetFiles,
file_id: str,
) -> EditorTaskResponse:
return await self._build_media_task(
project=project,
file_record=file_record,
file_id=file_id,
default_key=VIDEO_DATA_KEY,
categories=MEDIA_OBJECT_CATEGORIES,
)
async def upsert_annotation(self, project_id: str, file_id: str, request: UpsertAnnotationRequest) -> UpsertAnnotationResponse:
project = await self._get_project_or_404(project_id)