Skip to content

dimimal/semantics_segmentation_of_urban_environments

Repository files navigation

Semantics Segmentation of Urban Environments

This is my undergraduate dissertation project. The goal of this thesis is to examine and compare the results from two variations of CNN Encode-Decode arhitectures using Self-Normalization technique along with CRF-RNN post processing unit. Due to visualize the results of the model properly a Visualizer based on CityscapesScripts has been implemented to visualize the results.

Cityscapes Dataset

Cityscapes

Dependencies

  • python 2.7
  • keras 2.1
  • tensorflow 1.4
  • scikit-learn 0.19
  • openCV 2.4
  • numpy 1.13
  • scipy 0.13
  • pyQt4 for the Visualizer

Run pip install -r requirements.txt to intall the dependencies

Arguments

train.py [-h] [-n NETWORK] [-trp TRAINPATH] [-vdp VALIDATIONPATH] [-tsp TESTPATH] [-bs BATCHSIZE] [-crf] [-w [WEIGHTS]] [-m [MODEL]] [-e EPOCHS]

Results

Input Image

Installation

Run make inside lib/crfasrnn_keras/src/cpp to build highdimfilter module. Create the npy data files for the data generator using denseExtraction.py.

Check the examples below to train your model.

Examples

Training

python train.py -n bdcnn -trp trainpath -vdp validationpath -tsp testpath -bs 4 -crf -e 20

Resume Training

python train.py -trp trainpath -vdp validationpath -tsp testpath -bs 4 -w weightspath -m modelpath -e 20

Acknowledgments

Releases

No releases published

Packages

No packages published

Languages