@mleonard9/vin-scanner 0.2.8 → 0.2.9

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (3) hide show
  1. package/ios/VisionCameraBarcodeScanner.m +23 -12
  2. package/ios/VisionCameraTextRecognition.m +27 -27
  3. package/package.json +1 -1
package/ios/VisionCameraBarcodeScanner.m CHANGED
@@ -5,7 +5,6 @@
5
5
  #import <VisionCamera/Frame.h>
6
6
  #import <VisionCamera/SharedArray.h>
7
7
  #import <React/RCTBridgeModule.h>
8
- #import <CoreVideo/CoreVideo.h>
9
8
  @import MLKitVision;
10
9
 
11
10
  @interface VisionCameraBarcodeScannerPlugin : FrameProcessorPlugin
@@ -59,14 +58,30 @@
59
58
  orientation = [self orientationFromValue:orientationOverride fallback:orientation];
60
59
  }
61
60
 
62
- // Get image dimensions for coordinate transformation
63
- CVImageBufferRef imageBuffer = CMSampleBufferGetImageBuffer(buffer);
64
- size_t imageHeight = CVPixelBufferGetHeight(imageBuffer);
61
+ // Fix vertical flip issue on iOS by rotating orientation 180 degrees
62
+ // This ensures ML Kit processes the image in the correct orientation
63
+ UIImageOrientation correctedOrientation = orientation;
64
+ switch (orientation) {
65
+ case UIImageOrientationUp:
66
+ correctedOrientation = UIImageOrientationDown;
67
+ break;
68
+ case UIImageOrientationDown:
69
+ correctedOrientation = UIImageOrientationUp;
70
+ break;
71
+ case UIImageOrientationLeft:
72
+ correctedOrientation = UIImageOrientationRight;
73
+ break;
74
+ case UIImageOrientationRight:
75
+ correctedOrientation = UIImageOrientationLeft;
76
+ break;
77
+ default:
78
+ break;
79
+ }
65
80
 
66
81
  // VisionCamera already normalizes orientation per https://react-native-vision-camera.com/docs/guides/orientation,
67
82
  // so ML Kit just needs the frame's orientation metadata instead of rotating pixels manually.
68
83
  MLKVisionImage *image = [[MLKVisionImage alloc] initWithBuffer:buffer];
69
- image.orientation = orientation;
84
+ image.orientation = correctedOrientation;
70
85
  NSMutableArray *detections = [NSMutableArray array];
71
86
  __block NSDictionary *resultPayload = @{};
72
87
  dispatch_group_t dispatchGroup = dispatch_group_create();
@@ -97,13 +112,9 @@
97
112
  if (boxData != nil) {
98
113
  CGRect frameRect = barcode.frame;
99
114
  const NSUInteger baseIndex = idx * 6;
100
- // Transform Y coordinates to fix vertical flip issue on iOS
101
- // ML Kit returns coordinates in image space, but VisionCamera expects them flipped
102
- float minY = CGRectGetMinY(frameRect);
103
- float maxY = CGRectGetMaxY(frameRect);
104
- float imageHeightFloat = (float)imageHeight;
105
- boxData[baseIndex] = imageHeightFloat - maxY; // top = height - bottom
106
- boxData[baseIndex + 1] = imageHeightFloat - minY; // bottom = height - top
115
+ // Coordinates are now correct after orientation fix
116
+ boxData[baseIndex] = CGRectGetMinY(frameRect);
117
+ boxData[baseIndex + 1] = CGRectGetMaxY(frameRect);
107
118
  boxData[baseIndex + 2] = CGRectGetMinX(frameRect);
108
119
  boxData[baseIndex + 3] = CGRectGetMaxX(frameRect);
109
120
  boxData[baseIndex + 4] = CGRectGetWidth(frameRect);
package/ios/VisionCameraTextRecognition.m CHANGED
@@ -9,7 +9,6 @@
9
9
  #import <VisionCamera/VisionCameraProxyHolder.h>
10
10
  #import <VisionCamera/Frame.h>
11
11
  #import <VisionCamera/SharedArray.h>
12
- #import <CoreVideo/CoreVideo.h>
13
12
  @import MLKitVision;
14
13
 
15
14
  @interface VisionCameraTextRecognitionPlugin : FrameProcessorPlugin
@@ -88,13 +87,28 @@
88
87
  orientation = [self orientationFromValue:orientationOverride fallback:orientation];
89
88
  }
90
89
 
91
- // Get image dimensions for coordinate transformation
92
- CVImageBufferRef imageBuffer = CMSampleBufferGetImageBuffer(buffer);
93
- size_t imageHeight = CVPixelBufferGetHeight(imageBuffer);
94
- float imageHeightFloat = (float)imageHeight;
90
+ // Fix vertical flip issue on iOS by rotating orientation 180 degrees
91
+ // This ensures ML Kit processes the image in the correct orientation
92
+ UIImageOrientation correctedOrientation = orientation;
93
+ switch (orientation) {
94
+ case UIImageOrientationUp:
95
+ correctedOrientation = UIImageOrientationDown;
96
+ break;
97
+ case UIImageOrientationDown:
98
+ correctedOrientation = UIImageOrientationUp;
99
+ break;
100
+ case UIImageOrientationLeft:
101
+ correctedOrientation = UIImageOrientationRight;
102
+ break;
103
+ case UIImageOrientationRight:
104
+ correctedOrientation = UIImageOrientationLeft;
105
+ break;
106
+ default:
107
+ break;
108
+ }
95
109
 
96
110
  MLKVisionImage *image = [[MLKVisionImage alloc] initWithBuffer:buffer];
97
- image.orientation = orientation;
111
+ image.orientation = correctedOrientation;
98
112
  NSMutableArray *detections = [NSMutableArray array];
99
113
  NSMutableArray<NSArray<NSNumber *> *> *boxValues = [NSMutableArray array];
100
114
  NSString *language = arguments[@"language"] ?: self.preferredLanguage ?: @"latin";
@@ -112,11 +126,7 @@
112
126
  NSString *resultText = result.text;
113
127
  for (MLKTextBlock *block in result.blocks) {
114
128
  CGRect blockFrame = block.frame;
115
- // Transform Y coordinates to fix vertical flip issue on iOS
116
- float blockMinY = CGRectGetMinY(blockFrame);
117
- float blockMaxY = CGRectGetMaxY(blockFrame);
118
- float blockTransformedMinY = imageHeightFloat - blockMaxY;
119
- float blockTransformedMaxY = imageHeightFloat - blockMinY;
129
+ // Coordinates are now correct after orientation fix
120
130
 
121
131
  if (block.lines.count == 0) {
122
132
  NSMutableDictionary *entry = [[NSMutableDictionary alloc] init];
@@ -125,7 +135,7 @@
125
135
  entry[@"boxIndex"] = @(boxValues.count);
126
136
  [detections addObject:entry];
127
137
  [boxValues addObject:@[
128
- @(blockTransformedMinY), @(blockTransformedMaxY),
138
+ @(CGRectGetMinY(blockFrame)), @(CGRectGetMaxY(blockFrame)),
129
139
  @(CGRectGetMinX(blockFrame)), @(CGRectGetMaxX(blockFrame)),
130
140
  @(-1.f), @(-1.f), @(-1.f), @(-1.f),
131
141
  @(-1.f), @(-1.f), @(-1.f), @(-1.f)
@@ -133,11 +143,6 @@
133
143
  }
134
144
  for (MLKTextLine *line in block.lines) {
135
145
  CGRect lineFrame = line.frame;
136
- // Transform Y coordinates for line frame
137
- float lineMinY = CGRectGetMinY(lineFrame);
138
- float lineMaxY = CGRectGetMaxY(lineFrame);
139
- float lineTransformedMinY = imageHeightFloat - lineMaxY;
140
- float lineTransformedMaxY = imageHeightFloat - lineMinY;
141
146
 
142
147
  if (line.elements.count == 0) {
143
148
  NSMutableDictionary *entry = [[NSMutableDictionary alloc] init];
@@ -147,20 +152,15 @@
147
152
  entry[@"boxIndex"] = @(boxValues.count);
148
153
  [detections addObject:entry];
149
154
  [boxValues addObject:@[
150
- @(blockTransformedMinY), @(blockTransformedMaxY),
155
+ @(CGRectGetMinY(blockFrame)), @(CGRectGetMaxY(blockFrame)),
151
156
  @(CGRectGetMinX(blockFrame)), @(CGRectGetMaxX(blockFrame)),
152
- @(lineTransformedMinY), @(lineTransformedMaxY),
157
+ @(CGRectGetMinY(lineFrame)), @(CGRectGetMaxY(lineFrame)),
153
158
  @(CGRectGetMinX(lineFrame)), @(CGRectGetMaxX(lineFrame)),
154
159
  @(-1.f), @(-1.f), @(-1.f), @(-1.f)
155
160
  ]];
156
161
  }
157
162
  for (MLKTextElement *element in line.elements) {
158
163
  CGRect elementFrame = element.frame;
159
- // Transform Y coordinates for element frame
160
- float elementMinY = CGRectGetMinY(elementFrame);
161
- float elementMaxY = CGRectGetMaxY(elementFrame);
162
- float elementTransformedMinY = imageHeightFloat - elementMaxY;
163
- float elementTransformedMaxY = imageHeightFloat - elementMinY;
164
164
 
165
165
  NSMutableDictionary *entry = [[NSMutableDictionary alloc] init];
166
166
  entry[@"resultText"] = resultText ?: (id)kCFNull;
@@ -170,11 +170,11 @@
170
170
  entry[@"boxIndex"] = @(boxValues.count);
171
171
  [detections addObject:entry];
172
172
  [boxValues addObject:@[
173
- @(blockTransformedMinY), @(blockTransformedMaxY),
173
+ @(CGRectGetMinY(blockFrame)), @(CGRectGetMaxY(blockFrame)),
174
174
  @(CGRectGetMinX(blockFrame)), @(CGRectGetMaxX(blockFrame)),
175
- @(lineTransformedMinY), @(lineTransformedMaxY),
175
+ @(CGRectGetMinY(lineFrame)), @(CGRectGetMaxY(lineFrame)),
176
176
  @(CGRectGetMinX(lineFrame)), @(CGRectGetMaxX(lineFrame)),
177
- @(elementTransformedMinY), @(elementTransformedMaxY),
177
+ @(CGRectGetMinY(elementFrame)), @(CGRectGetMaxY(elementFrame)),
178
178
  @(CGRectGetMinX(elementFrame)), @(CGRectGetMaxX(elementFrame))
179
179
  ]];
180
180
  }
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@mleonard9/vin-scanner",
3
- "version": "0.2.8",
3
+ "version": "0.2.9",
4
4
  "description": "High-performance VIN scanner for React Native Vision Camera powered by Google ML Kit barcode + text recognition.",
5
5
  "main": "lib/commonjs/index",
6
6
  "module": "lib/module/index",