Dilated Continuous Random Field for Semantic Segmentation

Mean field approximation methodology has laid the foundation of modern Continuous Random Field (CRF) based solutions for the refinement of semantic segmentation. In this paper, we propose to relax the hard constraint of mean field approximation - minimizing the energy term of each node from probabilistic graphical model, by a global optimization with the proposed dilated sparse convolution module (DSConv). In addition, adaptive global average-pooling and adaptive global max-pooling are implemented as replacements of fully connected layers. In order to integrate DSConv, we design an end-to-end, time-efficient DilatedCRF pipeline. The unary energy term is derived either from pre-softmax and post-softmax features, or the predicted affordance map using a conventional classifier, making it easier to implement DilatedCRF for varieties of classifiers. We also present superior experimental results of proposed approach on the suction dataset comparing to other CRF-based approaches.

Authors

• Xi Mo
• Xiangyu Chen 0008
• Cuncong Zhong
• Rui Li
• Kaidong Li
• Usman Sajid

Keywords

• Graphical models
• Automation
• Convolution
• Affordances
• Semantics
• Pipelines
• Probabilistic logic
• Semantic Segmentation
• Continuous Field
• Continuous Random Field
• Global Optimization
• Mean-field
• Conventional Classification
• Probabilistic Graphical Models
• Neural Network
• Training Set
• Convolutional Network
• Convolutional Neural Network
• Negative Samples
• Recurrent Neural Network
• Gibbs Free Energy
• Global Features
• Intersection Over Union
• Trainable Parameters
• Separate Regions
• Inference Procedure
• Linear Layer
• Maximum A Posteriori
• Pairwise Potential
• Gated Recurrent Unit
• Pairwise Energy
• Global Terms
• Superpixel Segmentation
• Dice Score
• Pairwise Terms
• Image Segmentation