Traffic Light Recognition using Convolutional Neural Networks: A Survey

TL;DR

This survey provides a comprehensive overview of CNN-based traffic light recognition methods, categorizing them into three groups, analyzing datasets, and highlighting research gaps for autonomous driving applications.

Contribution

It offers a structured classification of CNN architectures for traffic light recognition and discusses datasets and research gaps, which was lacking in prior work.

Findings

CNN modifications improve detection accuracy

Multi-stage approaches combine rule-based and CNN methods

Identified key research gaps in datasets and model architectures

Abstract

Real-time traffic light recognition is essential for autonomous driving. Yet, a cohesive overview of the underlying model architectures for this task is currently missing. In this work, we conduct a comprehensive survey and analysis of traffic light recognition methods that use convolutional neural networks (CNNs). We focus on two essential aspects: datasets and CNN architectures. Based on an underlying architecture, we cluster methods into three major groups: (1) modifications of generic object detectors which compensate for specific task characteristics, (2) multi-stage approaches involving both rule-based and CNN components, and (3) task-specific single-stage methods. We describe the most important works in each cluster, discuss the usage of the datasets, and identify research gaps.