From Big to Small: Multi-Scale Local Planar Guidance for Monocular Depth Estimation

TL;DR

This paper introduces a multi-scale local planar guidance approach within an encoder-decoder neural network to improve monocular depth estimation, achieving superior results on benchmark datasets.

Contribution

It proposes a novel local planar guidance layer integrated at multiple decoding stages, enhancing depth prediction accuracy over existing methods.

Findings

Outperforms state-of-the-art methods on benchmark datasets

Significant improvement demonstrated through ablation studies

Effective multi-scale guidance enhances depth estimation quality

Abstract

Estimating accurate depth from a single image is challenging because it is an ill-posed problem as infinitely many 3D scenes can be projected to the same 2D scene. However, recent works based on deep convolutional neural networks show great progress with plausible results. The convolutional neural networks are generally composed of two parts: an encoder for dense feature extraction and a decoder for predicting the desired depth. In the encoder-decoder schemes, repeated strided convolution and spatial pooling layers lower the spatial resolution of transitional outputs, and several techniques such as skip connections or multi-layer deconvolutional networks are adopted to recover the original resolution for effective dense prediction. In this paper, for more effective guidance of densely encoded features to the desired depth prediction, we propose a network architecture that utilizes novel local planar guidance layers located at multiple stages in the decoding phase. We show that the proposed method outperforms the state-of-the-art works with significant margin evaluating on challenging benchmarks. We also provide results from an ablation study to validate the effectiveness of the proposed method.