Update Bounding Box and MLkit dependency

[mat2718]

Hi! 👋

Firstly, thanks for your work on this project! 🙂

Today I used patch-package to patch vision-camera-face-detector@0.1.8 for the project I'm working on.

fixes

• return normal bounding box parameters
• upgrade MLkit to work with other plugins such as OCR

i can also include the function im using to adjust the bounding box based on on the screen size the camera takes up. i use that function when using a header since the bounding box can get a little skewed.

Here is the diff that solved my problem:

diff --git a/node_modules/vision-camera-face-detector/android/build.gradle b/node_modules/vision-camera-face-detector/android/build.gradle
index 55bfaac..8ced7fb 100644
--- a/node_modules/vision-camera-face-detector/android/build.gradle
+++ b/node_modules/vision-camera-face-detector/android/build.gradle
@@ -56,6 +56,6 @@ dependencies {
     implementation "com.facebook.react:react-native:+"  // From node_modules
     implementation 'androidx.annotation:annotation:1.1.0'
     api project(":react-native-vision-camera")
-    implementation "androidx.camera:camera-core:1.1.0-alpha06"
-    implementation 'com.google.mlkit:face-detection:16.1.2'
+    implementation "androidx.camera:camera-core:1.1.0-alpha08"
+    implementation 'com.google.mlkit:face-detection:16.1.5'
 }
diff --git a/node_modules/vision-camera-face-detector/android/src/main/java/com/visioncamerafacedetector/VisionCameraFaceDetectorPlugin.java b/node_modules/vision-camera-face-detector/android/src/main/java/com/visioncamerafacedetector/VisionCameraFaceDetectorPlugin.java
index 3e4b3ff..18e0371 100644
--- a/node_modules/vision-camera-face-detector/android/src/main/java/com/visioncamerafacedetector/VisionCameraFaceDetectorPlugin.java
+++ b/node_modules/vision-camera-face-detector/android/src/main/java/com/visioncamerafacedetector/VisionCameraFaceDetectorPlugin.java
@@ -1,4 +1,4 @@

 import static java.lang.Math.ceil;
@@ -47,26 +47,13 @@ public class VisionCameraFaceDetectorPlugin extends FrameProcessorPlugin {
   private WritableMap processBoundingBox(Rect boundingBox) {
     WritableMap bounds = Arguments.createMap();

-    // Calculate offset (we need to center the overlay on the target)
-    Double offsetX =  (boundingBox.exactCenterX() - ceil(boundingBox.width())) / 2.0f;
-    Double offsetY =  (boundingBox.exactCenterY() - ceil(boundingBox.height())) / 2.0f;
-
-    Double x = boundingBox.right + offsetX;
-    Double y = boundingBox.top + offsetY;
-
-
-
-    bounds.putDouble("x", boundingBox.centerX() + (boundingBox.centerX() - x));
-    bounds.putDouble("y", boundingBox.centerY() + (y - boundingBox.centerY()));
+    bounds.putDouble("x", boundingBox.centerX());
+    bounds.putDouble("y", boundingBox.centerY());
+    bounds.putDouble("top", boundingBox.top);
+    bounds.putDouble("left", boundingBox.left);
     bounds.putDouble("width", boundingBox.width());
     bounds.putDouble("height", boundingBox.height());

-
-    bounds.putDouble("boundingCenterX", boundingBox.centerX());
-    bounds.putDouble("boundingCenterY", boundingBox.centerY());
-    bounds.putDouble("boundingExactCenterX", boundingBox.exactCenterX());
-    bounds.putDouble("boundingExactCenterY", boundingBox.exactCenterY());
-
     return bounds;
   }

diff --git a/node_modules/vision-camera-face-detector/src/index.ts b/node_modules/vision-camera-face-detector/src/index.ts
index 8ca44fb..abe1fd7 100644
--- a/node_modules/vision-camera-face-detector/src/index.ts
+++ b/node_modules/vision-camera-face-detector/src/index.ts
@@ -13,11 +13,14 @@ export interface Face {
   rightEyeOpenProbability: number;
   smilingProbability: number;
   bounds: {
-    y: number;
     x: number;
+    y: number;
+    top: number;
+    left: number;
     height: number;
     width: number;
   };
+
   contours: {
     FACE: Point[];
     NOSE_BOTTOM: Point[];

This issue body was partially generated by patch-package.

vision-camera-face-detector+0.1.8.patch

ts file to adjust bounding box based on view and frame

adjustToView.ts

export type Dimensions = {width: number; height: number};

// matches return bounds from custom face detection module
export type Rect = {
  top: number;
  left: number;
  height: number;
  width: number;
};

/**
 * It takes a frame and a view, and returns an object with two functions: adjustPoint and adjustRect
 * @param {Dimensions} frame - Dimensions - the dimensions of the video frame
 * @param {Dimensions} view - Dimensions
 * @returns An object with two functions.
 * @resource https://github.com/bglgwyng/FrameProcessorExample/blob/e8e99d58c878d4dce9a8adf74a7447d253be93ab/adjustToView.ts#L21
 */
const adjustToView = (
  frame: Dimensions,
  view: Dimensions,
  landscape: boolean,
  verticalCropPadding: number,
  horizontalCropPadding: number,
) => {
  'worklet';
  const {width, height} = view;
  /* Calculating the aspect ratio of the view. */
  const aspectRatio = width / height;

  const frameWidth = frame.width;
  const frameHeight = frame.height;

  /* Setting the widthRatio, heightRatio, offsetX, and offsetY to 0. */
  let widthRatio: number;
  let heightRatio: number;
  let offsetX = 0;
  let offsetY = 0;

  /* Calculating the ratio of the frame to the view. */

  const croppedFrameWidth = aspectRatio * frameHeight;
  const croppedFrameHeight = aspectRatio * frameWidth;
  if (!landscape) {
    offsetX = (frameWidth - croppedFrameWidth) / 2;
  } else {
    offsetY = (frameHeight - croppedFrameHeight) / 2;
  }
  heightRatio = height / frameHeight;
  widthRatio = width / croppedFrameWidth;

  /* Returning an object with two functions. */
  return {
    adjustPoint: (point: {x: number; y: number}) => ({
      x: (point.x - offsetX) * widthRatio,
      y: (point.y - offsetY) * heightRatio,
    }),
    adjustRect: (rect: Rect) => ({
      top: (rect.top - offsetY - verticalCropPadding) * heightRatio,
      left: (rect.left - offsetX - horizontalCropPadding) * widthRatio,
      height: (rect.height + verticalCropPadding) * heightRatio,
      width: (rect.width + horizontalCropPadding) * widthRatio,
    }),
  };
};

export default adjustToView;