Keywords: machine-learning, deep-learning, classification, android-implement, computer-vision, convolutional neural network, transfer learning.
This repository offers tutorials about TensorFlow on Android. It includes detailed steps to set up the environment, prepare data, train the model and optimize it.
TensorFlow is an open source library for numerical computation, specializing in machine learning applications. This project aims to use TensorFlow for object classification on Android.
- High level and low level APIs for data manipulation.
- Multi-GPU support.
- Training across distributed resources.
- Queues for placing data loading and preprocessing operations on the graph.
- Visualizing the graph using TensorBoard.
- Interactive logging events with TensorBoard.
- Model checkpointing.
- Comparable performance and GPU memory usage to Theano and CUDNN.
- Getting started with TensorFlow: https://www.tensorflow.org/get_started/
- Getting started with TensorFlow Android: https://www.tensorflow.org/mobile/android_build
- Additional resources: https://bit.ly/2Ikuwnt (Slide for android developer)
Follow the instructions on setting up the environment, preparing data, creating your first app, training and optimizing model, adding model files to the project, viewing the demo, and accessing the full source code!
Source code can be found here: https://bit.ly/2ILVkfv
Links to relevant resources and tutorials are included for further learning.
Any questions or issues are welcome!
Created by lilyu18 ❤️
Happy training!