Revisiting Dilated Convolution: A Simple Approach for Weakly- and Semi- Supervised Semantic Segmentation

Abstract

Despite the remarkable progress, weakly supervised segmentation approaches are still inferior to their fully supervised counterparts. We obverse the performance gap mainly comes from their limitation on learning to produce high-quality dense object localization maps from image-level supervision. To mitigate such a gap, we revisit the dilated convolution [1] and reveal how it can be utilized in a novel way to effectively overcome this critical limitation of weakly supervised segmentation approaches. Specifically, we find that varying dilation rates can effectively enlarge the receptive fields of convolutional kernels and more importantly transfer the surrounding discriminative information to non-discriminative object regions, promoting the emergence of these regions in the object localization maps. Then, we design a generic classification network equipped with convolutional blocks of different dilated rates. It can produce dense and reliable object localization maps and effectively benefit both weakly- and semi- supervised semantic segmentation. Despite the apparent simplicity, our proposed approach obtains superior performance over state-of-the-arts. In particular, it achieves 60.8% and 67.6% mIoU scores on Pascal VOC 2012 test set in weakly- (only image-level labels are available) and semi- (1,464 segmentation masks are available) supervised settings, which are the new state-of-the-arts.

arXiv commentAccepted by CVPR 2018 as Spotlight

References in corpus (18)

• Caffe: Convolutional Architecture for Fast Feature Embedding
• Conditional Random Fields as Recurrent Neural Networks
• CNN: Single-label to Multi-label
• STC: A Simple to Complex Framework for Weakly-supervised Semantic Segmentation
• Multiscale Combinatorial Grouping for Image Segmentation and Object Proposal Generation
• Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation
• Fully Convolutional Multi-Class Multiple Instance Learning
• Adversarial Complementary Learning for Weakly Supervised Object Localization
• Visual Translation Embedding Network for Visual Relation Detection
• Incorporating Network Built-in Priors in Weakly-supervised Semantic Segmentation
• Two-Phase Learning for Weakly Supervised Object Localization
• Weakly Supervised Semantic Segmentation using Web-Crawled Videos
• Built-in Foreground/Background Prior for Weakly-Supervised Semantic Segmentation
• Deep Self-Taught Learning for Weakly Supervised Object Localization
• Computational Baby Learning
• Temporal Dynamic Graph LSTM for Action-driven Video Object Detection
• Self-explanatory Deep Salient Object Detection
• Constrained Convolutional Neural Networks for Weakly Supervised Segmentation

Cited by in corpus (14)

• Devil in the Details: Towards Accurate Single and Multiple Human Parsing
• Rectifying Pseudo Label Learning via Uncertainty Estimation for Domain Adaptive Semantic Segmentation
• Reducing Information Bottleneck for Weakly Supervised Semantic Segmentation
• Self-Erasing Network for Integral Object Attention
• Shallow Feature Matters for Weakly Supervised Object Localization
• Weak-shot Semantic Segmentation by Transferring Semantic Affinity and Boundary
• Mask Based Unsupervised Content Transfer
• GuidedMix-Net: Learning to Improve Pseudo Masks Using Labeled Images as Reference
• Few-Shot Semantic Segmentation Augmented with Image-Level Weak Annotations
• Multi-Miner: Object-Adaptive Region Mining for Weakly-Supervised Semantic Segmentation
• Alleviating Semantic-level Shift: A Semi-supervised Domain Adaptation Method for Semantic Segmentation
• Class Activation Map generation by Multiple Level Class Grouping and Orthogonal Constraint
• TraceCaps: A Capsule-based Neural Network for Semantic Segmentation
• What Synthesis is Missing: Depth Adaptation Integrated with Weak Supervision for Indoor Scene Parsing