[Feature request] 32-bit face-detection native libraries are still built with 4 KB page alignment (not 16 KB)

[arvismobile]

ENVIRONMENT
Phone make and model: Pixel 9 Pro XL
OS version:
Google Play services version (only Android):
ML Kit APIs and versions:

STEPS
Build an Android app including ML Kit Face Detection (com.google.mlkit:face-detection:16.1.7).
Inspect the generated APK or AAB using: check_elf_alignment.sh app_debug.apk

EXPECTED
All native libraries should be 16 KB page-aligned.

ACTUAL
The arm64-v8a and x86_64 versions of libface_detector_v2_jni.so are correctly aligned to 16 KB. However, the armeabi-v7a and x86 variants remain aligned to 4 KB, indicating they are still produced via the older build pipeline.

Example (verified output): lib/x86_64/libface_detector_v2_jni.so: ALIGNED (214) lib/arm64-v8a/libface_detector_v2_jni.so: ALIGNED (214) lib/x86/libface_detector_v2_jni.so: UNALIGNED (212) lib/armeabi-v7a/libface_detector_v2_jni.so: UNALIGNED (212)

ADDITIONAL INFORMATION
This issue appears to originate from how the 32-bit variants of face-detection is published.

### Can you explain why this difference occurs?

[FaceAISDK]

libimage_processing_util_jni.so？？

The native library libimage_processing_util_jni.so included in your current ML Kit version is aligned to 4 KB, which causes errors on newer devices or during Google Play Console verification.

Here are the best ways to fix it, ranked from most recommended to least.

Solution 1: Add CameraX Dependency (Recommended)
This is the most effective workaround currently. Even if you don't use CameraX features, the CameraX 1.4.0+ library includes a newer, 16 KB-aligned version of libimage_processing_util_jni.so. Adding this dependency forces Gradle to use the fixed version of the .so file, overriding the old one from ML Kit.

Steps: Open your module-level build.gradle (usually app/build.gradle) and add the following line to your dependencies:

Gradle

dependencies {
    // ... other dependencies

    // Add CameraX core to get the 16KB-aligned native library
    implementation "androidx.camera:camera-core:1.4.0" 

    // Keep your existing ML Kit dependency
    implementation 'com.google.mlkit:face-detection:16.1.7'
}

Note: You must use version 1.4.2 or higher.

if your project dependency other lib that also dependency CameraX core and not 1.4.2 or higher.
you can force int app Gradle file like this

//CameraX，适配Android16 KB Page请强制升级到1.4.2 以上版本
def camera_version = "1.4.2"
configurations.configureEach {
    resolutionStrategy {
        force "androidx.camera:camera-core:$camera_version",
                "androidx.camera:camera-camera2:$camera_version",
                "androidx.camera:camera-lifecycle:$camera_version",
                "androidx.camera:camera-view:$camera_version"
    }
}

https://github.com/FaceAISDK/FaceAISDK_Android/blob/publish/demo/build.gradle

Good luck 🍀