The One Hundred Layers Tiramisu: Fully Convolutional DenseNets for Semantic Segmentation

TLDR

State‑of‑the‑art semantic segmentation relies on CNNs with a downsampling path, an upsampling path, and optional post‑processing such as CRFs, while DenseNets—whose dense layer connectivity improves accuracy and trainability—have recently excelled in image classification. We extend DenseNets to semantic segmentation by designing a fully convolutional architecture that preserves dense connectivity across downsampling and upsampling stages. The resulting model attains state‑of‑the‑art performance on CamVid and Gatech while using fewer parameters than current best entries and requiring no post‑processing or pretraining.

Abstract

State-of-the-art approaches for semantic image segmentation are built on Convolutional Neural Networks (CNNs). The typical segmentation architecture is composed of (a) a downsampling path responsible for extracting coarse semantic features, followed by (b) an upsampling path trained to recover the input image resolution at the output of the model and, optionally, (c) a post-processing module (e.g. Conditional Random Fields) to refine the model predictions. Recently, a new CNN architecture, Densely Connected Convolutional Networks (DenseNets), has shown excellent results on image classification tasks. The idea of DenseNets is based on the observation that if each layer is directly connected to every other layer in a feed-forward fashion then the network will be more accurate and easier to train. In this paper, we extend DenseNets to deal with the problem of semantic segmentation. We achieve state-of-the-art results on urban scene benchmark datasets such as CamVid and Gatech, without any further post-processing module nor pretraining. Moreover, due to smart construction of the model, our approach has much less parameters than currently published best entries for these datasets.