mcnDeepLab

This repo contains code to evaluate the models described in the paper:

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs 
Liang-Chieh Chen*, George Papandreou*, Iasonas Kokkinos, Kevin Murphy, and Alan L. Yuille (*equal contribution) 
Transactions on Pattern Analysis and Machine Intelligence (TPAMI)

The code is based on this caffe implementation.

The pre-trained models released with the caffe code which have been imported into matconvnet and can be downloaded here.

Demo

Running the deeplab_demo.m script will download a model trained on pascal voc 2012 data and run it on a sample image to produce the figure below:

Functionality

There is a script to evaluate trained models on the pascal voc 2012 dataset for semantic segmentation. The training code is still in the verification process.

Installation

This module can be installed with the following vl_contrib commands:

vl_contrib('install', 'mcnDeepLab') ;
vl_contrib('setup', 'mcnDeepLab') ;

The code has the following dependencies (these can similarly be added with vl_contrib):

• autonn - a wrapper module for matconvnet
• mcnExtraLayers - extra MatConvNet layers