Skip to content
A reimplementation of a fruit classifier in TensorFlow 2.0 based on the Fruit360 paper + dataset
Branch: master
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.idea Update SCM files Mar 18, 2019
data Fix imports Mar 18, 2019
model Add TensorBoard callback Mar 18, 2019
.gitattributes
.gitignore Update SCM files Mar 18, 2019
LICENSE Initial commit Mar 17, 2019
README.md
network-diagram.png Add information Mar 18, 2019
requirements.txt Add todo and get started sections Mar 18, 2019

README.md

FruitClassifier

A reimplementation of ArXiv paper 1712.00580 using TF 2.0's high level APIs.

Todo

  • Implement model saving
  • Implement test set metrics

Get Started

Requires TF 2.0 and Python 3.

  1. Download the Fruits360 dataset.
  2. Place Training and Test folders within data/.
  3. pip install -r requirements.txt
  4. cd model && python train.py
  5. Tensorboard logs will be under logs/.

TF 2.0 APIs Used

  • New tf.data API
    • lazy fetched dataset!
  • Improved tf.keras API

Network Structure

(pulled from the original paper)

Disclaimer

I did not write the paper or create the dataset. Most of the work was done by the paper authors. I implemented their network in TF 2.0 using the Keras API as a learning exercise.

You can’t perform that action at this time.