You've already forked DataMate
feat(annotation): 扩展标注模板的数据类型和标注类型选项
- 新增多种数据类型包括PDF、时间序列、聊天数据、HTML和表格数据 - 扩展分类标签体系增加音频语音、聊天评估、对话AI、生成式AI等分类 - 实现动态下拉选项渲染替代硬编码选项列表 - 添加超过100种新的标注类型涵盖音频语音、聊天评估、计算机视觉等领域 - 更新列表组件显示映射后的标签文本而非原始值 - 优化表单布局添加wrap属性支持选项换行显示
This commit is contained in:
@@ -17,6 +17,7 @@ import {
|
||||
updateAnnotationTemplateByIdUsingPut,
|
||||
} from "../annotation.api";
|
||||
import type { AnnotationTemplate } from "../annotation.model";
|
||||
import { DataTypeMap, ClassificationMap, AnnotationTypeMap } from "../annotation.const";
|
||||
import TagSelector from "./components/TagSelector";
|
||||
|
||||
const { TextArea } = Input;
|
||||
@@ -147,7 +148,7 @@ const TemplateForm: React.FC<TemplateFormProps> = ({
|
||||
/>
|
||||
</Form.Item>
|
||||
|
||||
<Space style={{ width: "100%" }} size="large">
|
||||
<Space style={{ width: "100%" }} size="large" wrap>
|
||||
<Form.Item
|
||||
label="数据类型"
|
||||
name="dataType"
|
||||
@@ -155,10 +156,9 @@ const TemplateForm: React.FC<TemplateFormProps> = ({
|
||||
style={{ width: 200 }}
|
||||
>
|
||||
<Select placeholder="选择数据类型">
|
||||
<Option value="image">图像</Option>
|
||||
<Option value="text">文本</Option>
|
||||
<Option value="audio">音频</Option>
|
||||
<Option value="video">视频</Option>
|
||||
{Object.entries(DataTypeMap).map(([key, item]) => (
|
||||
<Option key={key} value={item.value}>{item.label}</Option>
|
||||
))}
|
||||
</Select>
|
||||
</Form.Item>
|
||||
|
||||
@@ -168,12 +168,10 @@ const TemplateForm: React.FC<TemplateFormProps> = ({
|
||||
rules={[{ required: true, message: "请选择标注类型" }]}
|
||||
style={{ width: 220 }}
|
||||
>
|
||||
<Select placeholder="选择标注类型">
|
||||
<Option value="classification">分类</Option>
|
||||
<Option value="object-detection">目标检测</Option>
|
||||
<Option value="segmentation">分割</Option>
|
||||
<Option value="ner">命名实体识别</Option>
|
||||
<Option value="multi-stage">多阶段</Option>
|
||||
<Select placeholder="选择标注类型" showSearch optionFilterProp="children">
|
||||
{Object.entries(AnnotationTypeMap).map(([key, item]) => (
|
||||
<Option key={key} value={item.value}>{item.label}</Option>
|
||||
))}
|
||||
</Select>
|
||||
</Form.Item>
|
||||
|
||||
@@ -193,12 +191,10 @@ const TemplateForm: React.FC<TemplateFormProps> = ({
|
||||
name="category"
|
||||
style={{ width: 180 }}
|
||||
>
|
||||
<Select>
|
||||
<Option value="computer-vision">计算机视觉</Option>
|
||||
<Option value="nlp">自然语言处理</Option>
|
||||
<Option value="audio">音频</Option>
|
||||
<Option value="quality-control">质量控制</Option>
|
||||
<Option value="custom">自定义</Option>
|
||||
<Select showSearch optionFilterProp="children">
|
||||
{Object.entries(ClassificationMap).map(([key, item]) => (
|
||||
<Option key={key} value={item.value}>{item.label}</Option>
|
||||
))}
|
||||
</Select>
|
||||
</Form.Item>
|
||||
</Space>
|
||||
|
||||
@@ -112,10 +112,17 @@ const TemplateList: React.FC = () => {
|
||||
|
||||
const getCategoryColor = (category: string) => {
|
||||
const colors: Record<string, string> = {
|
||||
"audio-speech": "purple",
|
||||
"chat": "cyan",
|
||||
"computer-vision": "blue",
|
||||
"conversational-ai": "magenta",
|
||||
"generative-ai": "volcano",
|
||||
"nlp": "green",
|
||||
"audio": "purple",
|
||||
"quality-control": "orange",
|
||||
"ranking-scoring": "gold",
|
||||
"structured-data": "lime",
|
||||
"time-series": "geekblue",
|
||||
"video": "orange",
|
||||
"community": "pink",
|
||||
"custom": "default",
|
||||
};
|
||||
return colors[category] || "default";
|
||||
@@ -164,7 +171,7 @@ const TemplateList: React.FC = () => {
|
||||
key: "dataType",
|
||||
width: 120,
|
||||
render: (dataType: string) => (
|
||||
<Tag color="cyan">{dataType}</Tag>
|
||||
<Tag color="cyan">{DataTypeMap[dataType as keyof typeof DataTypeMap]?.label || dataType}</Tag>
|
||||
),
|
||||
},
|
||||
{
|
||||
@@ -173,7 +180,7 @@ const TemplateList: React.FC = () => {
|
||||
key: "labelingType",
|
||||
width: 150,
|
||||
render: (labelingType: string) => (
|
||||
<Tag color="geekblue">{labelingType}</Tag>
|
||||
<Tag color="geekblue">{AnnotationTypeMap[labelingType as keyof typeof AnnotationTypeMap]?.label || labelingType}</Tag>
|
||||
),
|
||||
},
|
||||
{
|
||||
@@ -182,7 +189,7 @@ const TemplateList: React.FC = () => {
|
||||
key: "category",
|
||||
width: 150,
|
||||
render: (category: string) => (
|
||||
<Tag color={getCategoryColor(category)}>{category}</Tag>
|
||||
<Tag color={getCategoryColor(category)}>{ClassificationMap[category as keyof typeof ClassificationMap]?.label || category}</Tag>
|
||||
),
|
||||
},
|
||||
{
|
||||
|
||||
@@ -84,24 +84,72 @@ export const DataTypeMap = {
|
||||
label: "视频",
|
||||
value: DataType.VIDEO
|
||||
},
|
||||
[DataType.PDF]: {
|
||||
label: "PDF文档",
|
||||
value: DataType.PDF
|
||||
},
|
||||
[DataType.TIMESERIES]: {
|
||||
label: "时间序列",
|
||||
value: DataType.TIMESERIES
|
||||
},
|
||||
[DataType.CHAT]: {
|
||||
label: "聊天数据",
|
||||
value: DataType.CHAT
|
||||
},
|
||||
[DataType.HTML]: {
|
||||
label: "HTML文档",
|
||||
value: DataType.HTML
|
||||
},
|
||||
[DataType.TABLE]: {
|
||||
label: "表格数据",
|
||||
value: DataType.TABLE
|
||||
},
|
||||
}
|
||||
|
||||
export const ClassificationMap = {
|
||||
[Classification.COMPUTER_VERSION]: {
|
||||
[Classification.AUDIO_SPEECH]: {
|
||||
label: "音频/语音处理",
|
||||
value: Classification.AUDIO_SPEECH
|
||||
},
|
||||
[Classification.CHAT]: {
|
||||
label: "聊天评估",
|
||||
value: Classification.CHAT
|
||||
},
|
||||
[Classification.COMPUTER_VISION]: {
|
||||
label: "计算机视觉",
|
||||
value: Classification.COMPUTER_VERSION
|
||||
value: Classification.COMPUTER_VISION
|
||||
},
|
||||
[Classification.CONVERSATIONAL_AI]: {
|
||||
label: "对话AI",
|
||||
value: Classification.CONVERSATIONAL_AI
|
||||
},
|
||||
[Classification.GENERATIVE_AI]: {
|
||||
label: "生成式AI",
|
||||
value: Classification.GENERATIVE_AI
|
||||
},
|
||||
[Classification.NLP]: {
|
||||
label: "自然语言处理",
|
||||
value: Classification.NLP
|
||||
},
|
||||
[Classification.AUDIO]: {
|
||||
label: "音频",
|
||||
value: Classification.AUDIO
|
||||
[Classification.RANKING_SCORING]: {
|
||||
label: "排名与评分",
|
||||
value: Classification.RANKING_SCORING
|
||||
},
|
||||
[Classification.QUALITY_CONTROL]: {
|
||||
label: "质量控制",
|
||||
value: Classification.QUALITY_CONTROL
|
||||
[Classification.STRUCTURED_DATA]: {
|
||||
label: "结构化数据解析",
|
||||
value: Classification.STRUCTURED_DATA
|
||||
},
|
||||
[Classification.TIME_SERIES]: {
|
||||
label: "时间序列分析",
|
||||
value: Classification.TIME_SERIES
|
||||
},
|
||||
[Classification.VIDEO]: {
|
||||
label: "视频处理",
|
||||
value: Classification.VIDEO
|
||||
},
|
||||
[Classification.COMMUNITY]: {
|
||||
label: "社区贡献",
|
||||
value: Classification.COMMUNITY
|
||||
},
|
||||
[Classification.CUSTOM]: {
|
||||
label: "自定义",
|
||||
@@ -110,22 +158,313 @@ export const ClassificationMap = {
|
||||
}
|
||||
|
||||
export const AnnotationTypeMap = {
|
||||
[AnnotationType.CLASSIFICATION]: {
|
||||
label: "分类",
|
||||
value: AnnotationType.CLASSIFICATION
|
||||
// ===== 音频/语音处理 =====
|
||||
[AnnotationType.ASR_SEGMENTS]: {
|
||||
label: "语音识别(分段)",
|
||||
value: AnnotationType.ASR_SEGMENTS
|
||||
},
|
||||
[AnnotationType.ASR]: {
|
||||
label: "语音识别",
|
||||
value: AnnotationType.ASR
|
||||
},
|
||||
[AnnotationType.CONVERSATION_ANALYSIS]: {
|
||||
label: "对话分析",
|
||||
value: AnnotationType.CONVERSATION_ANALYSIS
|
||||
},
|
||||
[AnnotationType.INTENT_CLASSIFICATION]: {
|
||||
label: "意图分类",
|
||||
value: AnnotationType.INTENT_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.SIGNAL_QUALITY]: {
|
||||
label: "信号质量检测",
|
||||
value: AnnotationType.SIGNAL_QUALITY
|
||||
},
|
||||
[AnnotationType.SOUND_EVENT_DETECTION]: {
|
||||
label: "声音事件检测",
|
||||
value: AnnotationType.SOUND_EVENT_DETECTION
|
||||
},
|
||||
[AnnotationType.SPEAKER_SEGMENTATION]: {
|
||||
label: "说话人分割",
|
||||
value: AnnotationType.SPEAKER_SEGMENTATION
|
||||
},
|
||||
[AnnotationType.SPEECH_TRANSCRIPTION]: {
|
||||
label: "语音转录",
|
||||
value: AnnotationType.SPEECH_TRANSCRIPTION
|
||||
},
|
||||
|
||||
// ===== 聊天评估 =====
|
||||
[AnnotationType.AGENT_FINE_TUNE]: {
|
||||
label: "Agent微调(无LLM)",
|
||||
value: AnnotationType.AGENT_FINE_TUNE
|
||||
},
|
||||
[AnnotationType.AGENT_FINE_TUNE_LLM]: {
|
||||
label: "Agent微调(有LLM)",
|
||||
value: AnnotationType.AGENT_FINE_TUNE_LLM
|
||||
},
|
||||
[AnnotationType.RED_TEAMING]: {
|
||||
label: "红队测试",
|
||||
value: AnnotationType.RED_TEAMING
|
||||
},
|
||||
[AnnotationType.RLHF_EVALUATION]: {
|
||||
label: "RLHF对话评估",
|
||||
value: AnnotationType.RLHF_EVALUATION
|
||||
},
|
||||
[AnnotationType.CHATBOT_EVALUATION]: {
|
||||
label: "聊天机器人评估",
|
||||
value: AnnotationType.CHATBOT_EVALUATION
|
||||
},
|
||||
|
||||
// ===== 计算机视觉 =====
|
||||
[AnnotationType.IMAGE_CAPTIONING]: {
|
||||
label: "图像描述",
|
||||
value: AnnotationType.IMAGE_CAPTIONING
|
||||
},
|
||||
[AnnotationType.IMAGE_CLASSIFICATION]: {
|
||||
label: "图像分类",
|
||||
value: AnnotationType.IMAGE_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.INVENTORY_TRACKING]: {
|
||||
label: "库存追踪",
|
||||
value: AnnotationType.INVENTORY_TRACKING
|
||||
},
|
||||
[AnnotationType.KEYPOINT_LABELING]: {
|
||||
label: "关键点标注",
|
||||
value: AnnotationType.KEYPOINT_LABELING
|
||||
},
|
||||
[AnnotationType.MEDICAL_IMAGE_CLASSIFICATION]: {
|
||||
label: "医学图像分类",
|
||||
value: AnnotationType.MEDICAL_IMAGE_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.MULTIPAGE_DOCUMENT]: {
|
||||
label: "多页文档标注",
|
||||
value: AnnotationType.MULTIPAGE_DOCUMENT
|
||||
},
|
||||
[AnnotationType.OBJECT_DETECTION]: {
|
||||
label: "目标检测",
|
||||
value: AnnotationType.OBJECT_DETECTION
|
||||
},
|
||||
[AnnotationType.SEGMENTATION]: {
|
||||
label: "分割",
|
||||
value: AnnotationType.SEGMENTATION
|
||||
[AnnotationType.OCR]: {
|
||||
label: "OCR识别",
|
||||
value: AnnotationType.OCR
|
||||
},
|
||||
[AnnotationType.PDF_OCR]: {
|
||||
label: "PDF OCR标注",
|
||||
value: AnnotationType.PDF_OCR
|
||||
},
|
||||
[AnnotationType.SEMANTIC_SEGMENTATION_MASK]: {
|
||||
label: "语义分割(掩码)",
|
||||
value: AnnotationType.SEMANTIC_SEGMENTATION_MASK
|
||||
},
|
||||
[AnnotationType.SEMANTIC_SEGMENTATION_POLYGON]: {
|
||||
label: "语义分割(多边形)",
|
||||
value: AnnotationType.SEMANTIC_SEGMENTATION_POLYGON
|
||||
},
|
||||
[AnnotationType.VISUAL_GENOME]: {
|
||||
label: "Visual Genome",
|
||||
value: AnnotationType.VISUAL_GENOME
|
||||
},
|
||||
[AnnotationType.VQA]: {
|
||||
label: "视觉问答",
|
||||
value: AnnotationType.VQA
|
||||
},
|
||||
|
||||
// ===== 对话AI =====
|
||||
[AnnotationType.COREFERENCE_RESOLUTION]: {
|
||||
label: "共指消解",
|
||||
value: AnnotationType.COREFERENCE_RESOLUTION
|
||||
},
|
||||
[AnnotationType.SLOT_FILLING]: {
|
||||
label: "槽填充",
|
||||
value: AnnotationType.SLOT_FILLING
|
||||
},
|
||||
[AnnotationType.RESPONSE_GENERATION]: {
|
||||
label: "响应生成",
|
||||
value: AnnotationType.RESPONSE_GENERATION
|
||||
},
|
||||
[AnnotationType.RESPONSE_SELECTION]: {
|
||||
label: "响应选择",
|
||||
value: AnnotationType.RESPONSE_SELECTION
|
||||
},
|
||||
|
||||
// ===== 生成式AI =====
|
||||
[AnnotationType.CHATBOT_ASSESSMENT]: {
|
||||
label: "聊天机器人评估",
|
||||
value: AnnotationType.CHATBOT_ASSESSMENT
|
||||
},
|
||||
[AnnotationType.RLHF_PREFERENCE]: {
|
||||
label: "RLHF人类偏好",
|
||||
value: AnnotationType.RLHF_PREFERENCE
|
||||
},
|
||||
[AnnotationType.LLM_RANKING]: {
|
||||
label: "LLM排名",
|
||||
value: AnnotationType.LLM_RANKING
|
||||
},
|
||||
[AnnotationType.LLM_GRADING]: {
|
||||
label: "LLM响应评分",
|
||||
value: AnnotationType.LLM_GRADING
|
||||
},
|
||||
[AnnotationType.SFT]: {
|
||||
label: "监督微调",
|
||||
value: AnnotationType.SFT
|
||||
},
|
||||
[AnnotationType.VISUAL_RANKING]: {
|
||||
label: "视觉排名",
|
||||
value: AnnotationType.VISUAL_RANKING
|
||||
},
|
||||
|
||||
// ===== 自然语言处理 =====
|
||||
[AnnotationType.CONTENT_MODERATION]: {
|
||||
label: "内容审核",
|
||||
value: AnnotationType.CONTENT_MODERATION
|
||||
},
|
||||
[AnnotationType.MACHINE_TRANSLATION]: {
|
||||
label: "机器翻译",
|
||||
value: AnnotationType.MACHINE_TRANSLATION
|
||||
},
|
||||
[AnnotationType.NER]: {
|
||||
label: "命名实体识别",
|
||||
value: AnnotationType.NER
|
||||
},
|
||||
[AnnotationType.QUESTION_ANSWERING]: {
|
||||
label: "问答",
|
||||
value: AnnotationType.QUESTION_ANSWERING
|
||||
},
|
||||
[AnnotationType.RELATION_EXTRACTION]: {
|
||||
label: "关系抽取",
|
||||
value: AnnotationType.RELATION_EXTRACTION
|
||||
},
|
||||
[AnnotationType.TAXONOMY]: {
|
||||
label: "层级分类",
|
||||
value: AnnotationType.TAXONOMY
|
||||
},
|
||||
[AnnotationType.TEXT_CLASSIFICATION]: {
|
||||
label: "文本分类",
|
||||
value: AnnotationType.TEXT_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.TEXT_SUMMARIZATION]: {
|
||||
label: "文本摘要",
|
||||
value: AnnotationType.TEXT_SUMMARIZATION
|
||||
},
|
||||
|
||||
// ===== 排名与评分 =====
|
||||
[AnnotationType.ASR_HYPOTHESES]: {
|
||||
label: "ASR假设选择",
|
||||
value: AnnotationType.ASR_HYPOTHESES
|
||||
},
|
||||
[AnnotationType.IMAGE_RETRIEVAL]: {
|
||||
label: "图像检索",
|
||||
value: AnnotationType.IMAGE_RETRIEVAL
|
||||
},
|
||||
[AnnotationType.DOCUMENT_RETRIEVAL]: {
|
||||
label: "文档检索",
|
||||
value: AnnotationType.DOCUMENT_RETRIEVAL
|
||||
},
|
||||
[AnnotationType.PAIRWISE_CLASSIFICATION]: {
|
||||
label: "成对分类",
|
||||
value: AnnotationType.PAIRWISE_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.PAIRWISE_REGRESSION]: {
|
||||
label: "成对回归",
|
||||
value: AnnotationType.PAIRWISE_REGRESSION
|
||||
},
|
||||
[AnnotationType.SERP_RANKING]: {
|
||||
label: "搜索排名",
|
||||
value: AnnotationType.SERP_RANKING
|
||||
},
|
||||
[AnnotationType.TEXT_TO_IMAGE]: {
|
||||
label: "文本生成图像",
|
||||
value: AnnotationType.TEXT_TO_IMAGE
|
||||
},
|
||||
|
||||
// ===== 结构化数据解析 =====
|
||||
[AnnotationType.FREEFORM_METADATA]: {
|
||||
label: "自由格式元数据",
|
||||
value: AnnotationType.FREEFORM_METADATA
|
||||
},
|
||||
[AnnotationType.HTML_ENTITY_RECOGNITION]: {
|
||||
label: "HTML实体识别",
|
||||
value: AnnotationType.HTML_ENTITY_RECOGNITION
|
||||
},
|
||||
[AnnotationType.PDF_CLASSIFICATION]: {
|
||||
label: "PDF分类",
|
||||
value: AnnotationType.PDF_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.TABULAR_DATA]: {
|
||||
label: "表格数据标注",
|
||||
value: AnnotationType.TABULAR_DATA
|
||||
},
|
||||
|
||||
// ===== 时间序列分析 =====
|
||||
[AnnotationType.ACTIVITY_RECOGNITION]: {
|
||||
label: "活动识别",
|
||||
value: AnnotationType.ACTIVITY_RECOGNITION
|
||||
},
|
||||
[AnnotationType.CHANGE_POINT_DETECTION]: {
|
||||
label: "变点检测",
|
||||
value: AnnotationType.CHANGE_POINT_DETECTION
|
||||
},
|
||||
[AnnotationType.ANOMALY_DETECTION]: {
|
||||
label: "异常检测",
|
||||
value: AnnotationType.ANOMALY_DETECTION
|
||||
},
|
||||
[AnnotationType.TIMESERIES_SIGNAL_QUALITY]: {
|
||||
label: "时序信号质量",
|
||||
value: AnnotationType.TIMESERIES_SIGNAL_QUALITY
|
||||
},
|
||||
[AnnotationType.TIMESERIES_FORECASTING]: {
|
||||
label: "时序预测",
|
||||
value: AnnotationType.TIMESERIES_FORECASTING
|
||||
},
|
||||
|
||||
// ===== 视频处理 =====
|
||||
[AnnotationType.VIDEO_CLASSIFICATION]: {
|
||||
label: "视频分类",
|
||||
value: AnnotationType.VIDEO_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.VIDEO_FRAME_CLASSIFICATION]: {
|
||||
label: "视频帧分类",
|
||||
value: AnnotationType.VIDEO_FRAME_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.VIDEO_OBJECT_TRACKING]: {
|
||||
label: "视频目标追踪",
|
||||
value: AnnotationType.VIDEO_OBJECT_TRACKING
|
||||
},
|
||||
[AnnotationType.VIDEO_TIMELINE_SEGMENTATION]: {
|
||||
label: "视频时间线分割",
|
||||
value: AnnotationType.VIDEO_TIMELINE_SEGMENTATION
|
||||
},
|
||||
|
||||
// ===== 社区贡献 =====
|
||||
[AnnotationType.MAMMOGRAM_CLASSIFICATION]: {
|
||||
label: "乳房X光分类",
|
||||
value: AnnotationType.MAMMOGRAM_CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.HTML_NER_TAGGING]: {
|
||||
label: "HTML NER标注",
|
||||
value: AnnotationType.HTML_NER_TAGGING
|
||||
},
|
||||
[AnnotationType.INVOICE_NER_BIO]: {
|
||||
label: "发票NER(BIO)",
|
||||
value: AnnotationType.INVOICE_NER_BIO
|
||||
},
|
||||
[AnnotationType.OCR_INVOICE_PRE_NER]: {
|
||||
label: "OCR发票预NER",
|
||||
value: AnnotationType.OCR_INVOICE_PRE_NER
|
||||
},
|
||||
[AnnotationType.TWITTER_SENTIMENT]: {
|
||||
label: "Twitter情感分析",
|
||||
value: AnnotationType.TWITTER_SENTIMENT
|
||||
},
|
||||
|
||||
// ===== 通用/遗留类型 =====
|
||||
[AnnotationType.CLASSIFICATION]: {
|
||||
label: "通用分类",
|
||||
value: AnnotationType.CLASSIFICATION
|
||||
},
|
||||
[AnnotationType.SEGMENTATION]: {
|
||||
label: "通用分割",
|
||||
value: AnnotationType.SEGMENTATION
|
||||
},
|
||||
}
|
||||
|
||||
export const TemplateTypeMap = {
|
||||
|
||||
@@ -79,26 +79,223 @@ export interface AnnotationTemplateListResponse {
|
||||
totalPages: number;
|
||||
}
|
||||
|
||||
/**
|
||||
* 数据类型枚举
|
||||
* 定义标注任务支持的数据类型
|
||||
*/
|
||||
export enum DataType {
|
||||
TEXT = "text",
|
||||
IMAGE = "image",
|
||||
AUDIO = "audio",
|
||||
VIDEO = "video",
|
||||
PDF = "pdf",
|
||||
TIMESERIES = "timeseries",
|
||||
CHAT = "chat",
|
||||
HTML = "html",
|
||||
TABLE = "table",
|
||||
}
|
||||
|
||||
/**
|
||||
* 模板分类枚举
|
||||
* 基于 Label Studio 模板分类体系
|
||||
*/
|
||||
export enum Classification {
|
||||
COMPUTER_VERSION = "computer-vision",
|
||||
// 音频/语音处理
|
||||
AUDIO_SPEECH = "audio-speech",
|
||||
// 聊天评估
|
||||
CHAT = "chat",
|
||||
// 计算机视觉
|
||||
COMPUTER_VISION = "computer-vision",
|
||||
// 对话AI
|
||||
CONVERSATIONAL_AI = "conversational-ai",
|
||||
// 生成式AI
|
||||
GENERATIVE_AI = "generative-ai",
|
||||
// 自然语言处理
|
||||
NLP = "nlp",
|
||||
AUDIO = "audio",
|
||||
QUALITY_CONTROL = "quality-control",
|
||||
CUSTOM = "custom"
|
||||
// 排名与评分
|
||||
RANKING_SCORING = "ranking-scoring",
|
||||
// 结构化数据解析
|
||||
STRUCTURED_DATA = "structured-data",
|
||||
// 时间序列分析
|
||||
TIME_SERIES = "time-series",
|
||||
// 视频处理
|
||||
VIDEO = "video",
|
||||
// 社区贡献
|
||||
COMMUNITY = "community",
|
||||
// 自定义
|
||||
CUSTOM = "custom",
|
||||
}
|
||||
|
||||
/**
|
||||
* 标注类型枚举
|
||||
* 定义各种具体的标注任务类型
|
||||
*/
|
||||
export enum AnnotationType {
|
||||
CLASSIFICATION = "classification",
|
||||
// ===== 音频/语音处理 =====
|
||||
// 自动语音识别（分段）
|
||||
ASR_SEGMENTS = "asr-segments",
|
||||
// 自动语音识别
|
||||
ASR = "asr",
|
||||
// 对话分析
|
||||
CONVERSATION_ANALYSIS = "conversation-analysis",
|
||||
// 意图分类
|
||||
INTENT_CLASSIFICATION = "intent-classification",
|
||||
// 信号质量检测
|
||||
SIGNAL_QUALITY = "signal-quality",
|
||||
// 声音事件检测
|
||||
SOUND_EVENT_DETECTION = "sound-event-detection",
|
||||
// 说话人分割
|
||||
SPEAKER_SEGMENTATION = "speaker-segmentation",
|
||||
// 语音转录
|
||||
SPEECH_TRANSCRIPTION = "speech-transcription",
|
||||
|
||||
// ===== 聊天评估 =====
|
||||
// 无LLM的Agent微调
|
||||
AGENT_FINE_TUNE = "agent-fine-tune",
|
||||
// 有LLM的Agent微调
|
||||
AGENT_FINE_TUNE_LLM = "agent-fine-tune-llm",
|
||||
// 红队测试
|
||||
RED_TEAMING = "red-teaming",
|
||||
// RLHF生产对话评估
|
||||
RLHF_EVALUATION = "rlhf-evaluation",
|
||||
// 聊天机器人评估
|
||||
CHATBOT_EVALUATION = "chatbot-evaluation",
|
||||
|
||||
// ===== 计算机视觉 =====
|
||||
// 图像描述
|
||||
IMAGE_CAPTIONING = "image-captioning",
|
||||
// 图像分类
|
||||
IMAGE_CLASSIFICATION = "image-classification",
|
||||
// 库存追踪
|
||||
INVENTORY_TRACKING = "inventory-tracking",
|
||||
// 关键点标注
|
||||
KEYPOINT_LABELING = "keypoint-labeling",
|
||||
// 医学图像分类
|
||||
MEDICAL_IMAGE_CLASSIFICATION = "medical-image-classification",
|
||||
// 多页文档标注
|
||||
MULTIPAGE_DOCUMENT = "multipage-document",
|
||||
// 目标检测（边界框）
|
||||
OBJECT_DETECTION = "object-detection",
|
||||
// OCR识别
|
||||
OCR = "ocr",
|
||||
// PDF OCR标注
|
||||
PDF_OCR = "pdf-ocr",
|
||||
// 语义分割（掩码）
|
||||
SEMANTIC_SEGMENTATION_MASK = "semantic-segmentation-mask",
|
||||
// 语义分割（多边形）
|
||||
SEMANTIC_SEGMENTATION_POLYGON = "semantic-segmentation-polygon",
|
||||
// Visual Genome
|
||||
VISUAL_GENOME = "visual-genome",
|
||||
// 视觉问答
|
||||
VQA = "vqa",
|
||||
|
||||
// ===== 对话AI =====
|
||||
// 共指消解与实体链接
|
||||
COREFERENCE_RESOLUTION = "coreference-resolution",
|
||||
// 意图分类与槽填充
|
||||
SLOT_FILLING = "slot-filling",
|
||||
// 响应生成
|
||||
RESPONSE_GENERATION = "response-generation",
|
||||
// 响应选择
|
||||
RESPONSE_SELECTION = "response-selection",
|
||||
|
||||
// ===== 生成式AI =====
|
||||
// 聊天机器人模型评估
|
||||
CHATBOT_ASSESSMENT = "chatbot-assessment",
|
||||
// RLHF人类偏好收集
|
||||
RLHF_PREFERENCE = "rlhf-preference",
|
||||
// LLM排名
|
||||
LLM_RANKING = "llm-ranking",
|
||||
// LLM响应评分
|
||||
LLM_GRADING = "llm-grading",
|
||||
// 监督微调
|
||||
SFT = "sft",
|
||||
// 视觉排名
|
||||
VISUAL_RANKING = "visual-ranking",
|
||||
|
||||
// ===== 自然语言处理 =====
|
||||
// 内容审核
|
||||
CONTENT_MODERATION = "content-moderation",
|
||||
// 机器翻译
|
||||
MACHINE_TRANSLATION = "machine-translation",
|
||||
// 命名实体识别
|
||||
NER = "ner",
|
||||
// 问答
|
||||
QUESTION_ANSWERING = "question-answering",
|
||||
// 关系抽取
|
||||
RELATION_EXTRACTION = "relation-extraction",
|
||||
// 分类法/层级分类
|
||||
TAXONOMY = "taxonomy",
|
||||
// 文本分类
|
||||
TEXT_CLASSIFICATION = "text-classification",
|
||||
// 文本摘要
|
||||
TEXT_SUMMARIZATION = "text-summarization",
|
||||
|
||||
// ===== 排名与评分 =====
|
||||
// ASR假设选择
|
||||
ASR_HYPOTHESES = "asr-hypotheses",
|
||||
// 基于内容的图像检索
|
||||
IMAGE_RETRIEVAL = "image-retrieval",
|
||||
// 文档检索
|
||||
DOCUMENT_RETRIEVAL = "document-retrieval",
|
||||
// 成对分类
|
||||
PAIRWISE_CLASSIFICATION = "pairwise-classification",
|
||||
// 成对回归
|
||||
PAIRWISE_REGRESSION = "pairwise-regression",
|
||||
// 搜索页面排名
|
||||
SERP_RANKING = "serp-ranking",
|
||||
// 文本到图像生成
|
||||
TEXT_TO_IMAGE = "text-to-image",
|
||||
|
||||
// ===== 结构化数据解析 =====
|
||||
// 自由格式元数据
|
||||
FREEFORM_METADATA = "freeform-metadata",
|
||||
// HTML实体识别
|
||||
HTML_ENTITY_RECOGNITION = "html-entity-recognition",
|
||||
// PDF分类
|
||||
PDF_CLASSIFICATION = "pdf-classification",
|
||||
// 表格数据标注
|
||||
TABULAR_DATA = "tabular-data",
|
||||
|
||||
// ===== 时间序列分析 =====
|
||||
// 活动识别
|
||||
ACTIVITY_RECOGNITION = "activity-recognition",
|
||||
// 变点检测
|
||||
CHANGE_POINT_DETECTION = "change-point-detection",
|
||||
// 异常检测
|
||||
ANOMALY_DETECTION = "anomaly-detection",
|
||||
// 时间序列信号质量
|
||||
TIMESERIES_SIGNAL_QUALITY = "timeseries-signal-quality",
|
||||
// 时间序列预测
|
||||
TIMESERIES_FORECASTING = "timeseries-forecasting",
|
||||
|
||||
// ===== 视频处理 =====
|
||||
// 视频分类
|
||||
VIDEO_CLASSIFICATION = "video-classification",
|
||||
// 视频帧分类
|
||||
VIDEO_FRAME_CLASSIFICATION = "video-frame-classification",
|
||||
// 视频目标追踪
|
||||
VIDEO_OBJECT_TRACKING = "video-object-tracking",
|
||||
// 视频时间线分割
|
||||
VIDEO_TIMELINE_SEGMENTATION = "video-timeline-segmentation",
|
||||
|
||||
// ===== 社区贡献 =====
|
||||
// 乳腺癌乳房X光分类
|
||||
MAMMOGRAM_CLASSIFICATION = "mammogram-classification",
|
||||
// HTML NER标注
|
||||
HTML_NER_TAGGING = "html-ner-tagging",
|
||||
// 发票NER标注（BIO格式）
|
||||
INVOICE_NER_BIO = "invoice-ner-bio",
|
||||
// OCR发票预NER（BIO格式）
|
||||
OCR_INVOICE_PRE_NER = "ocr-invoice-pre-ner",
|
||||
// Twitter/X情感分析
|
||||
TWITTER_SENTIMENT = "twitter-sentiment",
|
||||
|
||||
// ===== 通用/遗留类型 =====
|
||||
// 通用分类
|
||||
CLASSIFICATION = "classification",
|
||||
// 通用分割
|
||||
SEGMENTATION = "segmentation",
|
||||
NER = "ner"
|
||||
}
|
||||
|
||||
export enum TemplateType {
|
||||
|
||||
Reference in New Issue
Block a user