https://github.com/donfuxx/pixbadger

An Android app with tensorflow machine learning module for image recognition.
https://github.com/donfuxx/pixbadger

android android-app android-application deep-learning image image-classification image-processing image-recognition kotlin kotlin-android livedata machine-learning mvvm-architecture recyclerview rxjava2 tensorflow tensorflow-experiments tensorflow-models viewmodel

Last synced: 26 days ago
JSON representation

An Android app with tensorflow machine learning module for image recognition.

• Host: GitHub
• URL: https://github.com/donfuxx/pixbadger
• Owner: donfuxx
• License: apache-2.0
• Created: 2018-10-28T11:16:42.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2019-01-19T12:55:40.000Z (almost 6 years ago)
• Last Synced: 2024-12-06T21:56:47.581Z (about 1 month ago)
• Topics: android, android-app, android-application, deep-learning, image, image-classification, image-processing, image-recognition, kotlin, kotlin-android, livedata, machine-learning, mvvm-architecture, recyclerview, rxjava2, tensorflow, tensorflow-experiments, tensorflow-models, viewmodel
• Language: Python
• Homepage:
• Size: 43.2 MB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# About the PixBadger Android Image Classification App

This app was originally forked from the **Google TensorFlow codelab** (https://github.com/googlecodelabs/tensorflow-for-poets-2) but now it has diverged into something else:

Most of the Python code related to training and optimizing the TensorFlow AI model is exactly like in Google's TensorFlow codelab. Just small modifications & some shell scripts for convenient retraining of the model.

Most of the changes are related to the actual Android code: Currently the PixBadger app would **search your device for jpg images** and then **classify them with tensorflow** instead of taking image frames from the Camera, like it was done in the original codelab. Last not least code was **converted from Java to Kotlin**. Then the app will display all classified images and classification results in real time in a recylerview.

The UX/UI is minimalistic at the moment as the focus here at the moment is on what is happening "under the hood" at the moment. If I ever decide to move this project in a production ready state I would most likely re-design and re-arrange quite a few things in the user interface and the user flow.

I practice here a lot of cool technologies: **Tensorflow, RxJava2, Viewmodel & LiveData** just to name a few.

**Work-in-progress**