Distilling Pixel-Wise Feature Similarities for Semantic Segmentation

Abstract

Among the neural network compression techniques, knowledge distillation is an effective one which forces a simpler student network to mimic the output of a larger teacher network. However, most of such model distillation methods focus on the image-level classification task. Directly adapting these methods to the task of semantic segmentation only brings marginal improvements. In this paper, we propose a simple, yet effective knowledge representation referred to as pixel-wise feature similarities (PFS) to tackle the challenging distillation problem of semantic segmentation. The developed PFS encodes spatial structural information for each pixel location of the high-level convolutional features, which helps guide the distillation process in an easier way. Furthermore, a novel weighted pixel-level soft prediction imitation approach is proposed to enable the student network to selectively mimic the teacher network's output, according to their pixel-wise knowledge-gaps. Extensive experiments are conducted on the challenging datasets of Pascal VOC 2012, ADE20K and Pascal Context. Our approach brings significant performance improvements compared to several strong baselines and achieves new state-of-the-art results.

References in corpus (16)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• Distilling the Knowledge in a Neural Network
• Rethinking Atrous Convolution for Semantic Image Segmentation
• FitNets: Hints for Thin Deep Nets
• Do Deep Nets Really Need to be Deep?
• ParseNet: Looking Wider to See Better
• Compressing Deep Convolutional Networks using Vector Quantization
• An Analysis of Deep Neural Network Models for Practical Applications
• Like What You Like: Knowledge Distill via Neuron Selectivity Transfer
• Net2Net: Accelerating Learning via Knowledge Transfer
• Non-local Neural Networks
• N2N Learning: Network to Network Compression via Policy Gradient Reinforcement Learning
• Do Deep Convolutional Nets Really Need to be Deep and Convolutional?
• Improving Fast Segmentation With Teacher-student Learning
• In Teacher We Trust: Learning Compressed Models for Pedestrian Detection
• Model Distillation with Knowledge Transfer from Face Classification to Alignment and Verification

Cited by in corpus (2)

• Double Similarity Distillation for Semantic Image Segmentation
• Be Your Own Best Competitor! Multi-Branched Adversarial Knowledge Transfer