Closed-Form Approximate CRF Training for Scalable Image Segmentation

Abstract

Abstract. We present LS-CRF, a new method for training cyclic Conditional Random Fields (CRFs) from large datasets that is inspired by classical closed-form expressions for the maximum likelihood parameters of a generative graphical model with tree topology. Training a CRF with LS-CRF requires only solving a set of independent regression problems, each of which can be solved efficiently in closed form or by an iterative solver. This makes LS-CRF orders of magnitude faster than classical CRF training based on probabilistic inference, and at the same time more flexible and easier to implement than other approximate techniques, such as pseudolikelihood or piecewise training. We apply LS-CRF to the task of semantic image segmentation, showing that it achieves on par accuracy to other training techniques at higher speed, thereby allowing efficient CRF training from very large training sets. For example, training a linearly parameterized pairwise CRF on 150,000 images requires less than one hour on a modern workstation.

Video Saliency Detection via Pairwise Interaction

“…Focusing on two aforementioned problems, we use least squares CRF training to abandon solving the unary function for the sake of avoiding extracting (weak) saliency maps. And we utilize least squares CRF (LS-CRF) [20] training to improve efficiency. The energy function is modified as…”

IsMo-GAN: Adversarial Learning for Monocular Non-Rigid 3D Reconstruction

“…In this section, we elaborate on the main datasets [17,30] used to train the OD-Net, Rec-Net and the discriminator. In Sec.…”

CrossNet: Latent Cross-Consistency for Unpaired Image Translation

“…In addition to our manually composed dataset of elephants, we quantify our method's results by employing the UT Zappos50K shoes dataset [24] which present photos with white backgrounds as well as on the HDSeg dataset [11] of Horses with ground truth segmentation masks. During training to segment shoes, we synthetically replaced the white background with natural scenery photos.…”