Self-supervised monocular depth estimation with large kernel attention and dynamic scene perception

Abstract

Self-supervised monocular depth estimation has emerged as a promising approach since it does not rely on labeled training data. Most methods combine convolution and Transformer to model long-distance dependencies to estimate depth accurately. However, Transformer treats 2D image features as 1D sequences, and positional encoding somewhat mitigates the loss of spatial information between different feature blocks, tending to overlook channel features, which limit the performance of depth estimation. In this paper, we propose a self-supervised monocular depth estimation network to get finer details. Specifically, we propose a decoder based on large kernel attention, which can model long-distance dependencies without compromising the two-dimension structure of features while maintaining feature channel adaptivity. In addition, we introduce a dynamic scene perception (DSP) module, which dynamically adjusts the receptive fields to capture more accurate depth discontinuities context information, thereby enhancing the network’s ability to process complex scenes. Besides, we introduce an up-sampling module to accurately recover the fine details in the depth map. Our method achieves highly competitive results on the KITTI dataset (AbsRel = 0.095, SqRel = 0.613, RMSElog = 0.169, $\delta$1 = 0.907), and shows great generalization performance on the challenging indoor dataset NYUv2 dataset.