Starain chen/clean autolabel data while training (#712)

* add logical to clean auto-label data while training * 1. remove clear data confirm logical 2. change logical to clean old data Co-authored-by: Alex Chen <68627897+yongbing-chen@users.noreply.github.com>
microsoft · Nov 6, 2020 · 32cfaea · 32cfaea
1 parent 282d557
commit 32cfaea
Show file tree

Hide file tree

Showing 3 changed files with 43 additions and 14 deletions.
diff --git a/src/common/localization/es-cl.ts b/src/common/localization/es-cl.ts
@@ -131,8 +131,8 @@ export const spanish: IAppStrings = {
         addName: "Agregar nombre de modelo ...",
         downloadJson: "Descargar archivo JSON",
         trainConfirm: {
-            title: "Etiquetas no revisadas todavía",
-            message: "Hay archivos recientemente etiquetados automáticamente que aún no ha revisado, ¿desea entrenar con esos archivos?"
+            title: "Etiquetas aún no revisadas",
+            message: "Hay archivos recién etiquetados automáticamente aún no revisados por usted, ¿quieres entrenar con esos archivos?"
         },
         errors: {
             electron: {

diff --git a/src/react/components/pages/train/trainPage.tsx b/src/react/components/pages/train/trainPage.tsx
@@ -10,7 +10,7 @@ import IProjectActions, * as projectActions from "../../../../redux/actions/proj
 import IApplicationActions, * as applicationActions from "../../../../redux/actions/applicationActions";
 import IAppTitleActions, * as appTitleActions from "../../../../redux/actions/appTitleActions";
 import {
-    IApplicationState, IConnection, IProject, IAppSettings, FieldType, IRecentModel, AssetLabelingState,
+    IApplicationState, IConnection, IProject, IAppSettings, FieldType, IRecentModel, AssetLabelingState, IAssetMetadata,
 } from "../../../../models/applicationState";
 import TrainChart from "./trainChart";
 import TrainPanel from "./trainPanel";
@@ -83,7 +83,6 @@ function mapDispatchToProps(dispatch) {
 export default class TrainPage extends React.Component<ITrainPageProps, ITrainPageState> {
     private appInsights: any = null;
     private notAdjustedLabelsConfirm: React.RefObject<Confirm> = React.createRef();
-
     constructor(props) {
         super(props);
 
@@ -316,12 +315,6 @@ export default class TrainPage extends React.Component<ITrainPageProps, ITrainPa
         });
 
         this.trainProcess().then(async (trainResult) => {
-            this.setState((prevState, props) => ({
-                isTraining: false,
-                trainMessage: this.getTrainMessage(trainResult),
-                currTrainRecord: this.getProjectTrainRecord(),
-                modelName: "",
-            }));
             const assets = Object.values(this.props.project.assets);
             const assetService = new AssetService(this.props.project);
             for (const asset of assets) {
@@ -333,6 +326,12 @@ export default class TrainPage extends React.Component<ITrainPageProps, ITrainPa
                     await assetService.save({ ...metadata });
                 }
             }
+            this.setState((prevState, props) => ({
+                isTraining: false,
+                trainMessage: this.getTrainMessage(trainResult),
+                currTrainRecord: this.getProjectTrainRecord(),
+                modelName: "",
+            }));
             // reset localStorage successful train process
             localStorage.setItem("trainPage_inputs", "{}");
         }).catch((err) => {
@@ -367,7 +366,7 @@ export default class TrainPage extends React.Component<ITrainPageProps, ITrainPa
                 showTrainingFailedWarning: true,
                 trainingFailedMessage: isOnPrem ? interpolate(strings.train.errors.electron.cantAccessFiles, { folderUri: this.state.inputtedLabelFolderURL }) :
                     error?.message !== undefined
-                    ? error.message : error,
+                        ? error.message : error,
             });
         }
     }
@@ -389,6 +388,7 @@ export default class TrainPage extends React.Component<ITrainPageProps, ITrainPa
             trainSourceURL = provider.sas;
             trainPrefix = this.props.project.folderPath ? this.props.project.folderPath : "";
         }
+        await this.cleanLabelData();
         const payload = {
             source: trainSourceURL,
             sourceFilter: {
@@ -411,6 +411,36 @@ export default class TrainPage extends React.Component<ITrainPageProps, ITrainPa
             ServiceHelper.handleServiceError(err);
         }
     }
+    private async cleanLabelData() {
+        const allAssets = { ...this.props.project.assets };
+        await Object.values(allAssets)
+            .filter(asset => asset.labelingState !== AssetLabelingState.Trained)
+            .forEachAsync(async (asset) => {
+                const assetMetadata: IAssetMetadata = { ...await this.props.actions.loadAssetMetadata(this.props.project, asset) };
+                if (assetMetadata.asset.labelingState === AssetLabelingState.ManuallyLabeled
+                    && assetMetadata.labelData?.labels?.findIndex(label => label.confidence
+                        || label.originValue
+                        || label.revised
+                        || label.value?.findIndex(item => item["confidence"]) < 0) < 0
+                ) {
+                    return;
+                }
+                assetMetadata.asset.labelingState = AssetLabelingState.ManuallyLabeled;
+                if (assetMetadata.labelData) {
+                    assetMetadata.labelData.labelingState = AssetLabelingState.ManuallyLabeled;
+                }
+
+                assetMetadata.labelData?.labels?.forEach((label) => {
+                    delete label.confidence;
+                    delete label.originValue;
+                    delete label.revised;
+                    label.value?.forEach(item => {
+                        delete item["confidence"];
+                    });
+                });
+                await this.props.actions.saveAssetMetadata(this.props.project,assetMetadata);
+            });
+    }
 
     private async getTrainStatus(operationLocation: string): Promise<any> {
         const timeoutPerFileInMs = 10000;  // 10 second for each file

diff --git a/src/services/assetService.ts b/src/services/assetService.ts
@@ -5,7 +5,7 @@ import _ from "lodash";
 import Guard from "../common/guard";
 import {
     IAsset, AssetType, IProject, IAssetMetadata, AssetState,
-    ILabelData, ILabel, AssetLabelingState
+    ILabelData, ILabel, AssetLabelingState, IFormRegion
 } from "../models/applicationState";
 import { AssetProviderFactory, IAssetProvider } from "../providers/storage/assetProviderFactory";
 import { StorageProviderFactory, IStorageProvider } from "../providers/storage/storageProviderFactory";
@@ -85,13 +85,12 @@ export class AssetService {
             return result;
         };
         const getLabelValues = (field: any) => {
-            return field.elements.map((path: string) => {
+            return field.elements.map((path: string):IFormRegion => {
                 const pathArr = path.split('/').slice(1);
                 const word = pathArr.reduce((obj: any, key: string) => obj[key], { ...predictResults.analyzeResult });
                 return {
                     page: field.page,
                     text: word.text || word.state,
-                    confidence: word.confidence,
                     boundingBoxes: [getBoundingBox(field.page, word.boundingBox)]
                 };
             });