lostinet / semantic-segmentation

label the pixels of a road in images using a Fully Convolutional Network (FCN).

Semantic Segmentation

Introduction

In this project, you'll label the pixels of a road in images using a Fully Convolutional Network (FCN).

Setup

Frameworks and Packages

Make sure you have the following is installed:

• Python 3
• TensorFlow
• NumPy
• SciPy

Dataset

Download the Kitti Road dataset from here. Extract the dataset in the data folder. This will create the folder data_road with all the training a test images.

Run

Run the following file to run the project in jupyter notebook:

FCN.ipynb

Environment

https://github.com/udacity/CarND-Object-Detection-Lab

name: carnd-advdl-odlab
channels:
    - https://conda.anaconda.org/menpo
    - conda-forge
dependencies:
    - python==3.6
    - numpy
    - matplotlib
    - jupyter
    - pillow
    - scipy
    - ffmpeg
    - imageio==2.1.2
    - pip:
        - moviepy
        - tensorflow-gpu==1.2

Note If running this in Jupyter Notebook system messages, such as those regarding test status, may appear in the terminal rather than the notebook.

Submission

3. Submit the following in a zip file.

• helper.py
• FCN.ipynb
• project_tests.py
• Newest inference images from runs folder (all images from the most recent run)

in my case, the most recent data is put into 1511221281.479869 folder.