I have been trying to run the following snippet of code for deepface for comparing two images containing a face. Python code: face_similarity.py
import base64
from io import BytesIO
from PIL import Image
import numpy as np
import cv2
from deepface import DeepFace
import traceback
import tensorflow as tf
import vitis_quantizer
def convert_to_tflite(model, tflite_filename):
# Convert the model to TFLite format
converter = tf.lite.TFLiteConverter.from_saved_model(model)
tflite_model = converter.convert()
# Save the TFLite model to a file
with open(tflite_filename, 'wb') as f:
f.write(tflite_model)
def calculate_deepface_similarity(ekyc_base64, attendance_base64):
if not ekyc_base64 or not attendance_base64:
return "image is missing"
try:
ekyc_image = _decode_base64_image(ekyc_base64)
attendance_image = _decode_base64_image(attendance_base64)
if ekyc_image is None or attendance_image is None:
return "invalid image data"
# Ensure both images are in portrait orientation
ekyc_image = _convert_to_portrait(ekyc_image)
attendance_image = _convert_to_portrait(attendance_image)
ekyc_image_np = np.array(ekyc_image)
attendance_image_np = np.array(attendance_image)
# Convert BGR to RGB (DeepFace requires RGB images)
ekyc_image_rgb = cv2.cvtColor(ekyc_image_np, cv2.COLOR_BGR2RGB)
attendance_image_rgb = cv2.cvtColor(attendance_image_np, cv2.COLOR_BGR2RGB)
# Perform face recognition using Facenet512 model
result = DeepFace.verify(ekyc_image_rgb, attendance_image_rgb, model_name='Facenet512')
# Check the structure of the result variable
print("Result:", result)
similarity_percentage = (1 - result["distance"]) * 100
return similarity_percentage
except Exception as e:
print("Error processing images:", e)
return str(traceback.format_exc())
def _decode_base64_image(base64_string):
try:
image_data = base64.b64decode(base64_string)
image = Image.open(BytesIO(image_data))
return image
except Exception as e:
print("Error decoding base64 image:", e)
return None
def _convert_to_portrait(image):
width, height = image.size
if width < height:
return image
else:
return image.transpose(Image.ROTATE_90)
def encode_image_to_base64(file_path):
try:
with open(file_path, "rb") as image_file:
# Read the image file in binary mode
image_binary = image_file.read()
# Encode the binary data to base64
encoded_string = base64.b64encode(image_binary).decode("utf-8")
return encoded_string
except Exception as e:
print(f"Error encoding image: {e}")
return None
if __name__ == "__main__":
# Example usage
ekyc_base64 = encode_image_to_base64("Screenshot 2023-11-26 220725.png")
attendance_base64 = encode_image_to_base64("Screenshot 2023-11-26 220806.png")
similarity_percentage = calculate_deepface_similarity(ekyc_base64, attendance_base64)
print("Similarity Percentage:", similarity_percentage)
However, I'm facing the following error:
Traceback (most recent call last):
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/app/face_similarity.py", line 36, in calculate_deepface_similarity
result = DeepFace.verify(ekyc_image_rgb, attendance_image_rgb, model_name='Facenet512')
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/deepface/DeepFace.py", line 169, in verify
img1_embedding_obj = represent(
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/deepface/DeepFace.py", line 639, in represent
model = build_model(model_name)
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/deepface/DeepFace.py", line 80, in build_model
model = model()
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/deepface/basemodels/Facenet512.py", line 11, in loadModel
model = Facenet.InceptionResNetV2(dimension=512)
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/deepface/basemodels/Facenet.py", line 49, in InceptionResNetV2
inputs = Input(shape=(160, 160, 3))
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/keras/engine/input_layer.py", line 270, in Input
input_layer = InputLayer(**input_layer_config)
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/keras/engine/input_layer.py", line 122, in _init_
input_tensor = backend.placeholder(
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/keras/backend.py", line 1054, in placeholder
x = array_ops.placeholder(dtype, shape=shape, name=name)
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/ops/array_ops.py", line 2718, in placeholder
return gen_array_ops.placeholder(dtype=dtype, shape=shape, name=name)
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/ops/gen_array_ops.py", line 6031, in placeholder
_, _, _op, _outputs = _op_def_library._apply_op_helper(
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/framework/op_def_library.py", line 740, in _apply_op_helper
op = g._create_op_internal(op_type_name, inputs, dtypes=None,
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/framework/func_graph.py", line 593, in _create_op_internal
return super(FuncGraph, self)._create_op_internal( # pylint: disable=protected-access
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/framework/ops.py", line 3307, in _create_op_internal
node_def = _NodeDef(op_type, name, attrs)
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/tensorflow_core/python/framework/ops.py", line 1569, in _NodeDef
node_def.attr[k].CopyFrom(v)
File "/data/data/rural.ekyc/files/chaquopy/AssetFinder/requirements/google/protobuf/internal/containers.py", line 70, in _getitem_
return self._values[key]
TypeError: list indices must be integers or slices, not str
My build.gradle:
import java.text.SimpleDateFormat
plugins {
id 'com.android.application'
id 'org.jetbrains.kotlin.android'
id 'kotlin-android'
id 'kotlin-parcelize'
id 'kotlin-kapt'
id 'dagger.hilt.android.plugin'
id 'androidx.navigation.safeargs.kotlin'
id 'com.google.gms.google-services'
id 'com.google.firebase.crashlytics'
id 'com.chaquo.python'
}
android {
namespace 'rural.ekyc'
compileSdk 33
defaultConfig {
applicationId "rural.ekyc"
minSdk 24
targetSdk 33
versionCode 1
versionName "1.0.0"
testInstrumentationRunner "androidx.test.runner.AndroidJUnitRunner"
python{
pip {
install "deepface==0.0.79"
install "numpy"
install "opencv-python"
}
buildPython "/usr/local/bin/python3"
}
ndk {
abiFilters "armeabi-v7a", "arm64-v8a", "x86", "x86_64"
}
sourceSets{
main{
python.srcDir "src/main/python"
}
}
}
applicationVariants.all { variant ->
variant.outputs.all {
def date = new SimpleDateFormat("dd_MM_yy").format(new Date())
outputFileName = "eKYC_APP_${date}_${variant.buildType.name}_${defaultConfig.versionName}.apk"
}
}
buildTypes {
release {
minifyEnabled false
proguardFiles getDefaultProguardFile('proguard-android-optimize.txt'), 'proguard-rules.pro'
}
/* debug {
minifyEnabled true
proguardFiles getDefaultProguardFile('proguard-android-optimize.txt'), 'proguard-rules.pro'
}*/
}
compileOptions {
sourceCompatibility JavaVersion.VERSION_17
targetCompatibility JavaVersion.VERSION_17
}
kotlinOptions {
jvmTarget = '17'
}
buildFeatures {
dataBinding false
viewBinding true
}
}
dependencies {
//Androidx
implementation 'androidx.core:core-ktx:1.10.1'
implementation 'androidx.appcompat:appcompat:1.6.1'
implementation 'androidx.navigation:navigation-fragment-ktx:2.5.3'
implementation 'androidx.navigation:navigation-ui-ktx:2.5.3'
implementation 'androidx.constraintlayout:constraintlayout:2.1.4'
implementation 'androidx.annotation:annotation:1.6.0'
testImplementation 'junit:junit:4.13.2'
androidTestImplementation 'androidx.test.ext:junit:1.1.5'
androidTestImplementation 'androidx.test.espresso:espresso-core:3.5.1'
//Google
implementation 'com.google.firebase:firebase-crashlytics:18.4.0'
implementation 'com.google.firebase:firebase-analytics:21.3.0'
implementation 'com.google.android.material:material:1.9.0'
implementation 'com.google.android.gms:play-services-vision:20.1.3'
//Intuitx
implementation 'com.intuit.sdp:sdp-android:1.0.6'
implementation 'com.intuit.ssp:ssp-android:1.0.6'
//Retrofit
implementation 'com.squareup.retrofit2:retrofit:2.9.0'
implementation 'com.squareup.retrofit2:converter-gson:2.9.0'
implementation "com.squareup.okhttp3:okhttp:4.9.0"
implementation "com.squareup.okhttp3:logging-interceptor:4.9.0"
//lifecycle
implementation 'androidx.lifecycle:lifecycle-viewmodel-ktx:2.6.1'
implementation 'androidx.lifecycle:lifecycle-livedata-ktx:2.6.1'
implementation 'androidx.lifecycle:lifecycle-service:2.6.1'
//Coroutines
implementation "org.jetbrains.kotlinx:kotlinx-coroutines-core:1.6.4"
implementation 'org.jetbrains.kotlinx:kotlinx-coroutines-android:1.6.4'
//Dragger Hilt
implementation "com.google.dagger:hilt-android:$hilt_version"
kapt "com.google.dagger:hilt-android-compiler:$hilt_version"
//Glide
implementation 'com.github.bumptech.glide:glide:4.13.2'
implementation 'com.github.bumptech.glide:annotations:4.13.2'
implementation 'com.github.bumptech.glide:okhttp3-integration:4.12.0'
kapt 'com.github.bumptech.glide:compiler:4.13.2'
// Room
implementation "androidx.room:room-runtime:2.5.2"
kapt "androidx.room:room-compiler:2.5.2"
// Kotlin Extensions and Coroutines support for Room
implementation "androidx.room:room-ktx:2.5.2"
implementation 'io.github.rupinderjeet:kprogresshud:1.0.0'
implementation 'com.github.f0ris.sweetalert:library:1.5.6'
implementation 'com.github.chivorns:smartmaterialspinner:1.5.0'
implementation 'com.fasterxml.jackson.dataformat:jackson-dataformat-xml:2.15.1'
implementation ('com.thoughtworks.xstream:xstream:1.4.7') {
exclude group: 'xmlpull', module: 'xmlpull'
}
implementation group: 'org.bouncycastle', name: 'bcprov-jdk16', version: '1.45'
implementation group: 'javax.xml.crypto', name: 'jsr105-api', version: '1.0.1'
implementation (group: 'org.apache.santuario', name: 'xmlsec', version: '2.0.3') {
exclude group: "org.codehaus.woodstox"
}
implementation group: 'javax.xml.stream', name: 'stax-api', version: '1.0-2'
implementation 'com.alimuzaffar.lib:pinentryedittext:2.0.6'
//bio-metric verification
implementation 'androidx.biometric:biometric:1.1.0'
// rootbeer
implementation 'com.scottyab:rootbeer-lib:0.1.0'
}
The pip freeze > requirements.txt output for my Windows virtual environment running in Python 3.10.9 is the following
absl-py==2.0.0
astunparse==1.6.3
beautifulsoup4==4.12.2
blinker==1.7.0
cachetools==5.3.2
certifi==2023.7.22
charset-normalizer==3.3.2
click==8.1.7
colorama==0.4.6
deepface==0.0.79
filelock==3.13.1
fire==0.5.0
Flask==3.0.0
flatbuffers==23.5.26
gast==0.5.4
gdown==4.7.1
google-auth==2.23.4
google-auth-oauthlib==1.0.0
google-pasta==0.2.0
grpcio==1.59.2
gunicorn==21.2.0
h5py==3.10.0
idna==3.4
itsdangerous==2.1.2
Jinja2==3.1.2
keras==2.14.0
libclang==16.0.6
Markdown==3.5.1
MarkupSafe==2.1.3
ml-dtypes==0.2.0
mtcnn==0.1.1
numpy==1.26.2
oauthlib==3.2.2
opencv-python==4.8.1.78
opt-einsum==3.3.0
packaging==23.2
pandas==2.1.3
Pillow==10.1.0
protobuf==4.25.0
pyasn1==0.5.0
pyasn1-modules==0.3.0
PySocks==1.7.1
python-dateutil==2.8.2
pytz==2023.3.post1
requests==2.31.0
requests-oauthlib==1.3.1
retina-face==0.0.13
rsa==4.9
six==1.16.0
soupsieve==2.5
tensorboard==2.14.1
tensorboard-data-server==0.7.2
tensorflow==2.14.0
tensorflow-estimator==2.14.0
tensorflow-intel==2.14.0
tensorflow-io-gcs-filesystem==0.31.0
termcolor==2.3.0
tqdm==4.66.1
typing_extensions==4.8.0
tzdata==2023.3
urllib3==2.1.0
Werkzeug==3.0.1
wrapt==1.14.1
Android Studio version: Android Studio Giraffe | 2022.3.1 Patch 2
Gradle version: 8.1.2
DeepFace version: 0.0.79
Chaquopy version: 14.0.2
The dependencies I have installed deepface, numpy and opencv-python. Also, is there a possibility of running commands to check the python versions running so as to duplicate that in a virtual environment on local system. The code for face_similarity.py works fine on my local system and also in a Windows virtual environment.
If anyone has an idea of what I can do to fix this I would be glad if they helped me. Thanks.
I tried installing various combinations of libraries in my build.gradle file including deepface, open-cv-python, and often got the error that opencv-python version was not found too, and I'm not able to duplicate my python local environment onto Chaquopy Java Kotlin Android Studio
The pip freeze output from your Windows virtual environment is irrelevant, because Chaquopy creates a separate virtual environment using the Android packages from its repository. The newest Tensorflow version in that repository is 2.1.0, which is incompatible with the current protobuf version, 4.25.
To work around this, add the following line to the pip block in your build.gradle file:
install "protobuf<4.25"