Joint Training of Generic CNN-CRF Models with Stochastic Optimization

TL;DR

This paper introduces a scalable, end-to-end CNN-CRF training framework using stochastic optimization, applicable to various architectures, and demonstrates its effectiveness in semantic labeling of depth images.

Contribution

It presents a novel joint stochastic optimization method for CNN-CRF models that is general, scalable, and easy to implement.

Findings

Outperforms competing techniques in semantic labeling tasks.

Applicable to arbitrary CNN and CRF architectures.

Efficient GPU parallelization and low memory usage.

Abstract

We propose a new CNN-CRF end-to-end learning framework, which is based on joint stochastic optimization with respect to both Convolutional Neural Network (CNN) and Conditional Random Field (CRF) parameters. While stochastic gradient descent is a standard technique for CNN training, it was not used for joint models so far. We show that our learning method is (i) general, i.e. it applies to arbitrary CNN and CRF architectures and potential functions; (ii) scalable, i.e. it has a low memory footprint and straightforwardly parallelizes on GPUs; (iii) easy in implementation. Additionally, the unified CNN-CRF optimization approach simplifies a potential hardware implementation. We empirically evaluate our method on the task of semantic labeling of body parts in depth images and show that it compares favorably to competing techniques.