CAM-Convs: Camera-Aware Multi-Scale Convolutions for Single-View Depth

TL;DR

This paper introduces camera-aware multi-scale convolutions that enable depth estimation networks to generalize across different camera models, reducing the need for new training data when camera parameters change.

Contribution

We propose a novel convolutional method that incorporates camera parameters, improving cross-camera generalization in single-view depth estimation.

Findings

Significant improvement in depth prediction accuracy across different cameras.

Outperforms state-of-the-art methods in cross-camera scenarios.

Enhances the robustness of depth estimation networks.

Abstract

Single-view depth estimation suffers from the problem that a network trained on images from one camera does not generalize to images taken with a different camera model. Thus, changing the camera model requires collecting an entirely new training dataset. In this work, we propose a new type of convolution that can take the camera parameters into account, thus allowing neural networks to learn calibration-aware patterns. Experiments confirm that this improves the generalization capabilities of depth prediction networks considerably, and clearly outperforms the state of the art when the train and test images are acquired with different cameras.