Neuromorphic Monocular Depth Estimation with Uncertainty Modeling

TL;DR

This paper introduces a deep learning approach for monocular depth estimation from event camera data, incorporating uncertainty modeling to identify reliable depth predictions.

Contribution

It presents a novel framework combining multiple event representations with uncertainty estimation methods, trained on synthetic data and fine-tuned on real sequences.

Findings

10 bin log-normal and 5 bin evidential models perform best across metrics

Uncertainty estimation effectively indicates pixels with reliable depth

Different event representations yield similar performance

Abstract

Event cameras offer distinct advantages over conventional frame-based sensors, including microsecond-level temporal resolution, high dynamic range, and low bandwidth. In this paper, we predict per-pixel depth distributions from monocular event streams using deep neural networks. We estimate uncertainty using Gaussian, log-normal, and evidential learning frameworks. We compare six event representations: spatio-temporal voxel grids with 1, 5, 10, and 20 temporal bins, the Compact Spatio-Temporal Representation (CSTR), and Time-Ordered Recent Event (TORE) volumes. Our U-Net-based models are trained on synthetic data and then fine-tuned on real sequences. We evaluate performance using absolute relative error, root mean squared error, and the area under the sparsification error. Quantitative results show that the representations perform similarly, while 10 bin log-normal and 5 bin evidential learning perform best across metrics. Our experiments demonstrate that uncertainty estimation can be successfully integrated into event-based monocular depth estimation, and be used to indicate pixels with reliable depth.