himanshurawlani / fully_convolutional_network

Building, training and deploying a fully convolutional neural network in TensorFlow

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Implementing a fully convolutional network (FCN) in TensorFlow

The code in this repository is developed and tested on Ubuntu 18.04.3 LTS with Python 3.6.7. Following are the packages required to setup the environment:

  • TensorFlow 2.x (tensorflow==2.0.0)
  • OpenCV (opencv-python==4.1.2.30)
  • Scikit learn (sklearn==0.21.3)
  • Numpy (numpy==1.16.2)

Please refer the blogpost here for a detailed explanation of the project. It covers the following topics:

  1. Building a fully convolutional network (FCN) in TensorFlow using Keras
  2. Downloading and splitting a sample dataset
  3. Creating a generator in Keras to load and process a batch of data in memory
  4. Training the network with variable batch dimensions
  5. Deploying the model using TensorFlow Serving

I'd love to have your suggestions and improvements to the repository, feel free to raise a GitHub issue for the same.

About

Building, training and deploying a fully convolutional neural network in TensorFlow

License:MIT License


Languages

Language:Jupyter Notebook 72.8%Language:Python 27.2%