fix: fix bug in ocrs, remove node_modules from cv example app (#122)

NorbertKlockiewicz · web-flow · commit 45e40a6621b6 · 2025-03-07T13:48:11.000+01:00
## Description
&lt;!-- Provide a concise and descriptive summary of the changes
implemented in this PR. --&gt;

### Type of change
- [x] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality to not work as expected)
- [ ] Documentation update (improves or adds clarity to existing
documentation)

### Tested on
- [x] iOS
- [x] Android

### Testing instructions
&lt;!-- Provide step-by-step instructions on how to test your changes.
Include setup details if necessary. --&gt;

### Screenshots
&lt;!-- Add screenshots here, if applicable --&gt;

### Related issues
&lt;!-- Link related issues here using #issue-number --&gt;

### Checklist
- [x] I have performed a self-review of my code
- [ ] I have commented my code, particularly in hard-to-understand areas
- [ ] I have updated the documentation accordingly
- [ ] My changes generate no new warnings

### Additional notes
&lt;!-- Include any additional information, assumptions, or context that
reviewers might need to understand this PR. --&gt;
diff --git a/android/src/main/java/com/swmansion/rnexecutorch/VerticalOCR.kt b/android/src/main/java/com/swmansion/rnexecutorch/VerticalOCR.kt
@@ -155,7 +155,7 @@ class VerticalOCR(reactContext: ReactApplicationContext) :
 
         resMap.putString("text", text)
         resMap.putArray("bbox", box.toWritableArray())
-        resMap.putDouble("confidence", confidenceScore)
+        resMap.putDouble("score", confidenceScore)
 
         predictions.pushMap(resMap)
       }
diff --git a/android/src/main/java/com/swmansion/rnexecutorch/models/ocr/RecognitionHandler.kt b/android/src/main/java/com/swmansion/rnexecutorch/models/ocr/RecognitionHandler.kt
@@ -104,7 +104,7 @@ class RecognitionHandler(
 
       resMap.putString("text", decodedTexts[0])
       resMap.putArray("bbox", box.toWritableArray())
-      resMap.putDouble("confidence", confidenceScore)
+      resMap.putDouble("score", confidenceScore)
 
       res.pushMap(resMap)
     }
diff --git a/examples/computer-vision/node_modules/react-native-executorch/ios/RnExecutorch/models/ocr/VerticalDetector.h b/examples/computer-vision/node_modules/react-native-executorch/ios/RnExecutorch/models/ocr/VerticalDetector.h
diff --git a/examples/computer-vision/node_modules/react-native-executorch/ios/RnExecutorch/models/ocr/VerticalDetector.mm b/examples/computer-vision/node_modules/react-native-executorch/ios/RnExecutorch/models/ocr/VerticalDetector.mm
diff --git a/examples/computer-vision/package.json b/examples/computer-vision/package.json
@@ -16,6 +16,7 @@
     "metro-config": "^0.81.0",
     "react": "18.3.1",
     "react-native": "0.76.3",
+    "react-native-audio-api": "^0.4.13",
     "react-native-executorch": "^0.3.0",
     "react-native-image-picker": "^7.2.2",
     "react-native-loading-spinner-overlay": "^3.0.1",
diff --git a/examples/computer-vision/yarn.lock b/examples/computer-vision/yarn.lock
@@ -3889,6 +3889,7 @@ __metadata:
     metro-config: ^0.81.0
     react: 18.3.1
     react-native: 0.76.3
+    react-native-audio-api: ^0.4.13
     react-native-executorch: ^0.3.0
     react-native-image-picker: ^7.2.2
     react-native-loading-spinner-overlay: ^3.0.1
@@ -7486,6 +7487,18 @@ __metadata:
   languageName: node
   linkType: hard
 
+"react-native-audio-api@npm:^0.4.13":
+  version: 0.4.13
+  resolution: "react-native-audio-api@npm:0.4.13"
+  peerDependencies:
+    react: "*"
+    react-native: "*"
+  bin:
+    setup-rn-audio-api-web: scripts/setup-rn-audio-api-web.js
+  checksum: 9a5db4626a0663224cdcdc957ed5176df9ed65df9f58bf06c5eae2a91895b7ee0f0a632ecaa79b23ef7536983485cea2720ca46201e2e690548f400b66970ff6
+  languageName: node
+  linkType: hard
+
 "react-native-executorch@npm:^0.3.0":
   version: 0.3.0
   resolution: "react-native-executorch@npm:0.3.0"
diff --git a/examples/llama/package.json b/examples/llama/package.json
@@ -16,6 +16,7 @@
     "metro-config": "^0.81.0",
     "react": "18.3.1",
     "react-native": "0.76.3",
+    "react-native-audio-api": "^0.4.13",
     "react-native-executorch": "^0.3.0",
     "react-native-loading-spinner-overlay": "^3.0.1",
     "react-native-markdown-display": "^7.0.2",
diff --git a/examples/llama/yarn.lock b/examples/llama/yarn.lock
@@ -6074,6 +6074,7 @@ __metadata:
     metro-config: ^0.81.0
     react: 18.3.1
     react-native: 0.76.3
+    react-native-audio-api: ^0.4.13
     react-native-executorch: ^0.3.0
     react-native-loading-spinner-overlay: ^3.0.1
     react-native-markdown-display: ^7.0.2
@@ -7555,6 +7556,18 @@ __metadata:
   languageName: node
   linkType: hard
 
+"react-native-audio-api@npm:^0.4.13":
+  version: 0.4.13
+  resolution: "react-native-audio-api@npm:0.4.13"
+  peerDependencies:
+    react: "*"
+    react-native: "*"
+  bin:
+    setup-rn-audio-api-web: scripts/setup-rn-audio-api-web.js
+  checksum: 9a5db4626a0663224cdcdc957ed5176df9ed65df9f58bf06c5eae2a91895b7ee0f0a632ecaa79b23ef7536983485cea2720ca46201e2e690548f400b66970ff6
+  languageName: node
+  linkType: hard
+
 "react-native-executorch@npm:^0.3.0":
   version: 0.3.0
   resolution: "react-native-executorch@npm:0.3.0"
diff --git a/ios/RnExecutorch/OCR.mm b/ios/RnExecutorch/OCR.mm
@@ -1,8 +1,8 @@
 #import "OCR.h"
 #import "models/ocr/Detector.h"
 #import "models/ocr/RecognitionHandler.h"
-#import "utils/ImageProcessor.h"
 #import "models/ocr/utils/Constants.h"
+#import "utils/ImageProcessor.h"
 #import <ExecutorchLib/ETModel.h>
 #import <React/RCTBridgeModule.h>
 
@@ -80,14 +80,11 @@ of different sizes (e.g. large - 512x64, medium - 256x64, small - 128x64).
   @try {
     cv::Mat image = [ImageProcessor readImage:input];
     NSArray *result = [detector runModel:image];
-    cv::Size detectorSize = [detector getModelImageSize];
-    const CGFloat recognizerRatio = recognizerImageSize / detectorSize.width;
     cv::cvtColor(image, image, cv::COLOR_BGR2GRAY);
-    result = [self->recognitionHandler
-            recognize:result
-              imgGray:image
-         desiredWidth:detectorSize.width * recognizerRatio
-        desiredHeight:detectorSize.height * recognizerRatio];
+    result = [self->recognitionHandler recognize:result
+                                         imgGray:image
+                                    desiredWidth:recognizerImageSize
+                                   desiredHeight:recognizerImageSize];
     resolve(result);
   } @catch (NSException *exception) {
     reject(@"forward_error",
diff --git a/ios/RnExecutorch/models/ocr/Detector.mm b/ios/RnExecutorch/models/ocr/Detector.mm
@@ -21,7 +21,7 @@ @implementation Detector {
   NSArray *inputShape = [module getInputShape:@0];
   NSNumber *widthNumber = inputShape[inputShape.count - 2];
   NSNumber *heightNumber = inputShape.lastObject;
-
+  
   const int height = [heightNumber intValue];
   const int width = [widthNumber intValue];
   modelSize = cv::Size(height, width);
diff --git a/ios/RnExecutorch/models/ocr/Recognizer.mm b/ios/RnExecutorch/models/ocr/Recognizer.mm
@@ -14,8 +14,8 @@ @implementation Recognizer {
 
 - (cv::Size)getModelImageSize {
   NSArray *inputShape = [module getInputShape:@0];
-  NSNumber *widthNumber = inputShape[inputShape.count - 2];
-  NSNumber *heightNumber = inputShape.lastObject;
+  NSNumber *widthNumber = inputShape.lastObject;
+  NSNumber *heightNumber = inputShape[inputShape.count - 2];
 
   const int height = [heightNumber intValue];
   const int width = [widthNumber intValue];
@@ -24,8 +24,8 @@ @implementation Recognizer {
 
 - (cv::Size)getModelOutputSize {
   NSArray *outputShape = [module getOutputShape:@0];
-  NSNumber *widthNumber = outputShape[outputShape.count - 2];
-  NSNumber *heightNumber = outputShape.lastObject;
+  NSNumber *widthNumber = outputShape.lastObject;
+  NSNumber *heightNumber = outputShape[outputShape.count - 2];
 
   const int height = [heightNumber intValue];
   const int width = [widthNumber intValue];
diff --git a/ios/RnExecutorch/models/ocr/utils/RecognizerUtils.mm b/ios/RnExecutorch/models/ocr/utils/RecognizerUtils.mm
@@ -251,17 +251,17 @@ + (double)computeConfidenceScore:(NSArray<NSNumber *> *)valuesArray
 
     points.emplace_back(cv::Point2f(point.x, point.y));
   }
-
+  
   cv::Rect rect = cv::boundingRect(points);
   cv::Mat croppedImage = img(rect);
+  cv::cvtColor(croppedImage, croppedImage, cv::COLOR_BGR2GRAY);
+  cv::resize(croppedImage, croppedImage, cv::Size(smallVerticalRecognizerWidth, recognizerHeight), 0, 0,
+               cv::INTER_AREA);
+  cv::medianBlur(img, img, 1);
   return croppedImage;
 }
 
 + (cv::Mat)cropSingleCharacter:(cv::Mat)img {
-  cv::cvtColor(img, img, cv::COLOR_BGR2GRAY);
-  cv::resize(img, img, cv::Size(smallVerticalRecognizerWidth, recognizerHeight), 0, 0,
-             cv::INTER_AREA);
-  cv::medianBlur(img, img, 1);
 
   cv::Mat histogram;
 

Original file line number	Diff line number	Diff line change
`@@ -155,7 +155,7 @@ class VerticalOCR(reactContext: ReactApplicationContext) :`
`155`	`155`
`156`	`156`	`resMap.putString("text", text)`
`157`	`157`	`resMap.putArray("bbox", box.toWritableArray())`
`158`		`- resMap.putDouble("confidence", confidenceScore)`
	`158`	`+ resMap.putDouble("score", confidenceScore)`
`159`	`159`
`160`	`160`	`predictions.pushMap(resMap)`
`161`	`161`	`}`
Original file line number	Diff line number	Diff line change
`@@ -104,7 +104,7 @@ class RecognitionHandler(`
`104`	`104`
`105`	`105`	`resMap.putString("text", decodedTexts[0])`
`106`	`106`	`resMap.putArray("bbox", box.toWritableArray())`
`107`		`- resMap.putDouble("confidence", confidenceScore)`
	`107`	`+ resMap.putDouble("score", confidenceScore)`
`108`	`108`
`109`	`109`	`res.pushMap(resMap)`
`110`	`110`	`}`