Delve into deploying DL models on Android using TensorFlow Lite. Gain insights into training, converting models, and practical applications through case studies for efficient mobile integration.

v4.tar.gz

tensorflow

tensorboard_keras

tensorboard_tf_function

CNN_on_CIFAR10

 tf_splits_slicing_api

image_augmentation

keras_preprocessing_api

tf_image_module

training_validation_testing

transfer_learning_keras

android

TFLite_Model_Maker

TF_object_detection_API

tensorflow-rb

TestingJob

tensorboard_tf_function-copy

Traditionally, computationally intensive deep learning (DL) models aren’t deployed to mobile devices due to their limited memory and computational power. But with a lightweight framework such as TensorFlow (TF) Lite, DL models can be deployed and run efficiently on mobile and edge devices, benefiting from fast response time, data privacy, and cost reduction in cloud computing resources. 

This course aims to equip you with the techniques to deploy DL models on Android devices using the TF Lite framework. You’ll start with a quick introduction to Python and machine learning. Next, you’ll explore different learning paradigms and DL models. You’ll also get practical knowledge of the applications of TF in the context of DL. The course covers TF Lite for mobile applications with different case studies showcasing its use in Android apps.

By the end of this course, you’ll gain the necessary skills to train DL models, convert them to the TF Lite format, and deploy them into your Android apps.

Deep Learning for Android Apps

import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt

# Load the TensorFlow Lite model into the interpreter
interpreter = tf.lite.Interpreter(model_path="TFLite_converted_model.tflite")   
interpreter.allocate_tensors()

# Get input and output tensors
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()

input_shape = input_details[0]['shape']
output_shape = output_details[0]['shape']

# Print name, shape, and type of model input and output
print("Input details",
     "\nName:", input_details[0]['name'],
      "\nShape:", input_details[0]['shape'],
      "\nType:", input_details[0]['dtype'],
      "\nIndex:", input_details[0]['index'])

print("\n\nOutput details",
      "name:", output_details[0]['name'],
      "\nshape:", output_details[0]['shape'],
       "\ntype:", output_details[0]['dtype'],
       "\nIndex:", output_details[0]['index'])

# Load a test image to be interpreted by the TF Lite model
#image_path = '/usr/local/notebooks/datasets/horses_or_humans_dataset/horse-or-human/horse-or-human/validation/humans/valhuman01-00.png'
image_path = '/usr/local/notebooks/datasets/horses_or_humans_dataset/horse-or-human/horse-or-human/validation/horses/horse2-011.png'
input_image = tf.keras.preprocessing.image.load_img(image_path, target_size=(224, 224))
input_image = tf.keras.preprocessing.image.img_to_array(input_image)
input_image = np.expand_dims(input_image, axis=0)
input_image = input_image / 255.0

# Set the input tensor
interpreter.set_tensor(input_details[0]['index'], input_image)

# Run inference
interpreter.invoke()

# Get the output tensor
output_data = interpreter.get_tensor(output_details[0]['index'])
print('\nOutput data:', output_data)
predicted_label_index = np.argmax(output_data)
print('\npredicted label index:', predicted_label_index)

# Display the image and predicted label: 0/1 for horse/human
plt.imshow(tf.squeeze(input_image))
plt.xticks([])
plt.yticks([])
plt.grid(False)
predicted_label = 'Human' if output_data[0,1] > output_data[0,0] else 'Horse'
plt.xlabel('Predicted label: %s' % predicted_label)
plt.savefig('output/test_image.png', dpi = 300)

Learn to apply the TF Lite Interpreter to classify images. 

TF Lite Interpreter (Part 2)

Learn to apply the TF Lite Interpreter to classify images.

Getting Started with Python

Machine Learning (ML) and Deep Learning (DL)

Customer Segmentation with K-Means Clustering

TensorFlow (TF)

Cats vs Dogs Classification with Convolutional Neural Networks

Dataset Processing Using TensorFlow

Keras: High-Level TF API

Diabetes Prediction Using Keras

Quick Start with Android Apps

TensorFlow (TF) Lite

Image Classification Apps Using TF Lite

Object Detection Apps Using TF Lite

Appendix

DL Model Using TF, Keras, and TF Lite

TF Lite Interpreter (Part 2)