shubham0204

// ML Model to detect the age of a person, hosted on Heroku
class AgeDetectionModel {

    private val herokuModelPredictURL = "https://age-detection-tf-app.herokuapp.com/predict"
    private val mediaType = "application/json".toMediaType()
    private val okHttpClient = OkHttpClient()

    interface PredictionCallback {
        fun onResult( age : Int )
        fun onError( error : String )

shubham0204

// Detect faces in the given image and crop them.
// Pass the cropped faces to the AgeDetectionModel
private fun detectFaces(image: Bitmap) {
    val inputImage = InputImage.fromBitmap(image, 0)
    firebaseFaceDetector.process(inputImage)
        .addOnSuccessListener { faces ->
            if ( faces.size != 0 ) {
                progressDialog.apply {
                    dismiss()
                    setMessage( "📍 Posting to server ...")

shubham0204

dependencies {
    implementation 'androidx.core:core-ktx:1.7.0'
    implementation 'androidx.appcompat:appcompat:1.4.1'
    implementation 'com.google.android.material:material:1.5.0'
    implementation 'androidx.constraintlayout:constraintlayout:2.1.3'

    // 1. MLKit Face Detection
    implementation 'com.google.android.gms:play-services-mlkit-face-detection:17.0.0'
    
    // 2. OkHttp is a HTTP client for Android

shubham0204

private lateinit var activityMainBinding : ActivityMainBinding
private lateinit var progressDialog : ProgressDialog

override fun onCreate(savedInstanceState: Bundle?) {
    super.onCreate(savedInstanceState)
    activityMainBinding = ActivityMainBinding.inflate( layoutInflater )
    setContentView( activityMainBinding.root )

    progressDialog = ProgressDialog( this )

shubham0204

from flask import Flask , jsonify , request
from PIL import Image
import tensorflow as tf
import numpy as np
import base64
import io

# Loading the Keras model to perform inference
# Download the model from this release ->
# https://github.com/shubham0204/Age-Gender_Estimation_TF-Android/releases/tag/v1.0

shubham0204

import math
import matplotlib.pyplot as plt

def poisson( lambda_ , x ):
    return ( (lambda_)**x * math.exp(-lambda_) ) / math.factorial(x)
  
x = []
y = []
for i in range( 100 ):
    x.append( i )

shubham0204

Blogs

• Explained: How Does L1 Regularization Perform Feature Selection?
• llama.cpp: Writing A Simple C++ Inference Program for GGUF LLM Models
• Building On-Device Face Recognition In Android
• From Python To Android: HF Sentence Transformers (Embeddings)
• On-Device Machine Learning In Android: Frameworks and Ecosystem
• Using C/C++ in Android: A Comprehensive Guide For Beginners
• [Building A Cross-Pl

shubham0204

hidden_dims = 128
token_mixing_mlp_dims = 64
channel_mixing_mlp_dims = 128
patch_size = 9
num_classes = 10
num_mixer_layers = 4
input_image_shape = ( 32 , 32 , 3 )

inputs = tf.keras.layers.Input( shape=input_image_shape )

shubham0204

hidden_dims = 128
token_mixing_mlp_dims = 64
channel_mixing_mlp_dims = 128
patch_size = 9
num_classes = 10
num_mixer_layers = 4
reshape_image_dim = 72
input_image_shape = ( 32 , 32 , 3 )

inputs = tf.keras.layers.Input( shape=input_image_shape )

shubham0204

# Mixer layer consisting of token mixing MLPs and channel mixing MLPs
# input shape -> ( batch_size , channels , num_patches )
# output shape -> ( batch_size , channels , num_patches )
def mixer( x , token_mixing_mlp_dims , channel_mixing_mlp_dims ):
    # inputs x of are of shape ( batch_size , num_patches , channels )
    # Note: "channels" is used instead of "embedding_dims"
    
    # Add token mixing MLPs
    token_mixing_out = token_mixing( x , token_mixing_mlp_dims )
    # Shape of token_mixing_out -> ( batch_size , channels , num_patches )