From ffedf2e961d1d17cdb8cb51b79b3616965bb5001 Mon Sep 17 00:00:00 2001
From: Abhinav <abhinavk.grd@gmail.com>
Date: Mon, 11 Apr 2022 14:15:55 +0530
Subject: [PATCH] move accurary filtering to tesseract Service

---
 .../machineLearning/tesseractService.ts       | 24 ++++++++++++-------
 src/services/machineLearning/textService.ts   | 18 +++++++-------
 src/types/machineLearning/index.ts            |  5 ++--
 3 files changed, 27 insertions(+), 20 deletions(-)

diff --git a/src/services/machineLearning/tesseractService.ts b/src/services/machineLearning/tesseractService.ts
index 172d42b85..877957fdf 100644
--- a/src/services/machineLearning/tesseractService.ts
+++ b/src/services/machineLearning/tesseractService.ts
@@ -7,20 +7,19 @@ import {
     Versioned,
 } from 'types/machineLearning';
 
-import Tesseract, { createWorker, RecognizeResult } from 'tesseract.js';
+import Tesseract, { createWorker } from 'tesseract.js';
 import QueueProcessor from 'services/queueProcessor';
 import { CustomError } from 'utils/error';
 import { imageBitmapToBlob, resizeToSquare } from 'utils/image';
 import { getFileType } from 'services/upload/readFileService';
 import { FILE_TYPE } from 'constants/file';
+import { makeID } from 'utils/user';
 
 class TesseractService implements TextDetectionService {
     private tesseractWorker: Tesseract.Worker;
     public method: Versioned<TextDetectionMethod>;
     private ready: Promise<void>;
-    private textDetector = new QueueProcessor<
-        Tesseract.RecognizeResult | Error
-    >(1);
+    private textDetector = new QueueProcessor<Tesseract.Word[] | Error>(1);
     public constructor() {
         this.method = {
             value: 'Tesseract',
@@ -59,8 +58,9 @@ class TesseractService implements TextDetectionService {
     }
 
     async detectText(
-        imageBitmap: ImageBitmap
-    ): Promise<RecognizeResult | Error> {
+        imageBitmap: ImageBitmap,
+        minAccuracy: number
+    ): Promise<Tesseract.Word[] | Error> {
         const response = this.textDetector.queueUpRequest(async () => {
             const imageHeight = Math.min(imageBitmap.width, imageBitmap.height);
             const imageWidth = Math.max(imageBitmap.width, imageBitmap.height);
@@ -102,9 +102,17 @@ class TesseractService implements TextDetectionService {
             }
 
             const tesseractWorker = await this.getTesseractWorker();
-            const detections = await tesseractWorker.recognize(file);
+            const id = makeID(6);
+            console.time('detecting text ' + id);
 
-            return detections;
+            const detections = await tesseractWorker.recognize(file);
+            console.timeEnd('detecting text ' + id);
+
+            const filteredDetections = detections.data.words.filter(
+                ({ confidence }) => confidence >= minAccuracy
+            );
+
+            return filteredDetections;
         });
         try {
             return await response.promise;
diff --git a/src/services/machineLearning/textService.ts b/src/services/machineLearning/textService.ts
index 16b76e64d..4a1d2eca0 100644
--- a/src/services/machineLearning/textService.ts
+++ b/src/services/machineLearning/textService.ts
@@ -34,24 +34,22 @@ class TextService {
             fileContext
         );
 
-        console.time('detecting text ' + fileContext.enteFile.id);
         const textDetections =
-            await syncContext.textDetectionService.detectText(imageBitmap);
-        console.timeEnd('detecting text ' + fileContext.enteFile.id);
+            await syncContext.textDetectionService.detectText(
+                imageBitmap,
+                syncContext.config.textDetection.minAccuracy
+            );
         if (textDetections instanceof Error) {
             newMlFile.errorCount = 2;
             newMlFile.lastErrorMessage = textDetections.message;
             return;
         }
-        const detectedText: DetectedText[] = textDetections.data.words
-            .filter(
-                ({ confidence }) =>
-                    confidence >= syncContext.config.textDetection.minAccuracy
-            )
-            .map(({ bbox, confidence, text }) => ({
+        const detectedText: DetectedText[] = textDetections.map(
+            ({ bbox, confidence, text }) => ({
                 fileID: fileContext.enteFile.id,
                 detection: { bbox, confidence, word: text.toLocaleLowerCase() },
-            }));
+            })
+        );
         newMlFile.text = detectedText;
         console.log(
             '[MLService] Detected text: ',
diff --git a/src/types/machineLearning/index.ts b/src/types/machineLearning/index.ts
index 8b54fac8a..826683894 100644
--- a/src/types/machineLearning/index.ts
+++ b/src/types/machineLearning/index.ts
@@ -395,8 +395,9 @@ export interface TextDetectionService {
     method: Versioned<TextDetectionMethod>;
     // init(): Promise<void>;
     detectText(
-        imageBitmap: ImageBitmap
-    ): Promise<Tesseract.RecognizeResult | Error>;
+        imageBitmap: ImageBitmap,
+        minAccuracy: number
+    ): Promise<Tesseract.Word[] | Error>;
     dispose(): Promise<void>;
 }