Boosting Deep Neural Networks with Geometrical Prior Knowledge: A Survey

TL;DR

This survey reviews methods of integrating geometrical prior knowledge into neural networks to improve data efficiency and interpretability, with potential applications in 3D object detection for autonomous driving.

Contribution

It provides a comprehensive overview of approaches for embedding geometrical priors into neural networks and discusses their relevance to 3D object detection.

Findings

Increased data efficiency through geometrical priors

Enhanced interpretability of neural network outputs

Potential improvements in autonomous driving applications

Abstract

Deep Neural Networks achieve state-of-the-art results in many different problem settings by exploiting vast amounts of training data. However, collecting, storing and - in the case of supervised learning - labelling the data is expensive and time-consuming. Additionally, assessing the networks' generalization abilities or predicting how the inferred output changes under input transformations is complicated since the networks are usually treated as a black box. Both of these problems can be mitigated by incorporating prior knowledge into the neural network. One promising approach, inspired by the success of convolutional neural networks in computer vision tasks, is to incorporate knowledge about symmetric geometrical transformations of the problem to solve that affect the output in a predictable way. This promises an increased data efficiency and more interpretable network outputs. In this survey, we try to give a concise overview about different approaches that incorporate geometrical prior knowledge into neural networks. Additionally, we connect those methods to 3D object detection for autonomous driving, where we expect promising results when applying those methods.