Rethinking Atrous Convolution for Semantic Image Segmentation

TLDR

The study revisits atrous convolution for semantic image segmentation, proposing an enhanced Atrous Spatial Pyramid Pooling module that incorporates image‑level global context to boost performance. The authors design cascaded and parallel atrous‑convolution modules with multiple rates to capture multi‑scale context, augmenting the ASPP module and training the system on the PASCAL VOC 2012 dataset. The resulting DeepLabv3 model outperforms earlier DeepLab versions without DenseCRF post‑processing and matches state‑of‑the‑art performance on PASCAL VOC 2012.

Abstract

In this work, we revisit atrous convolution, a powerful tool to explicitly adjust filter's field-of-view as well as control the resolution of feature responses computed by Deep Convolutional Neural Networks, in the application of semantic image segmentation. To handle the problem of segmenting objects at multiple scales, we design modules which employ atrous convolution in cascade or in parallel to capture multi-scale context by adopting multiple atrous rates. Furthermore, we propose to augment our previously proposed Atrous Spatial Pyramid Pooling module, which probes convolutional features at multiple scales, with image-level features encoding global context and further boost performance. We also elaborate on implementation details and share our experience on training our system. The proposed `DeepLabv3' system significantly improves over our previous DeepLab versions without DenseCRF post-processing and attains comparable performance with other state-of-art models on the PASCAL VOC 2012 semantic image segmentation benchmark.