Convolution kernel adaptation to calibrated fisheye

TL;DR

This paper proposes a method to adapt convolution kernels in CNNs to fisheye camera distortions using calibration data, enabling better performance in depth estimation and segmentation with minimal fine-tuning.

Contribution

It introduces a calibration-based kernel deformation technique that aligns CNN receptive fields with fisheye distortions, facilitating effective transfer from perspective datasets.

Findings

Improved depth estimation accuracy on fisheye images.

Enhanced semantic segmentation performance with kernel adaptation.

Requires only brief fine-tuning on small datasets.

Abstract

Convolution kernels are the basic structural component of convolutional neural networks (CNNs). In the last years there has been a growing interest in fisheye cameras for many applications. However, the radially symmetric projection model of these cameras produces high distortions that affect the performance of CNNs, especially when the field of view is very large. In this work, we tackle this problem by proposing a method that leverages the calibration of cameras to deform the convolution kernel accordingly and adapt to the distortion. That way, the receptive field of the convolution is similar to standard convolutions in perspective images, allowing us to take advantage of pre-trained networks in large perspective datasets. We show how, with just a brief fine-tuning stage in a small dataset, we improve the performance of the network for the calibrated fisheye with respect to standard convolutions in depth estimation and semantic segmentation.