Abstract:Understanding other drivers' intentions is crucial for safe driving. The role of taillights in conveying these intentions is underemphasized in current autonomous driving systems. Accurately identifying taillight signals is essential for predicting vehicle behavior and preventing collisions. Open-source taillight datasets are scarce, often small and inconsistently annotated. To address this gap, we introduce a new large-scale taillight dataset called TLD. Sourced globally, our dataset covers diverse traffic scenarios. To our knowledge, TLD is the first dataset to separately annotate brake lights and turn signals in real driving scenarios. We collected 17.78 hours of driving videos from the internet. This dataset consists of 152k labeled image frames sampled at a rate of 2 Hz, along with 1.5 million unlabeled frames interspersed throughout. Additionally, we have developed a two-stage vehicle light detection model consisting of two primary modules: a vehicle detector and a taillight classifier. Initially, YOLOv10 and DeepSORT captured consecutive vehicle images over time. Subsequently, the two classifiers work simultaneously to determine the states of the brake lights and turn signals. A post-processing procedure is then used to eliminate noise caused by misidentifications and provide the taillight states of the vehicle within a given time frame. Our method shows exceptional performance on our dataset, establishing a benchmark for vehicle taillight detection. The dataset is available at <a class="link-external link-https" href="https://huggingface.co/datasets/ChaiJohn/TLD/tree/main" rel="external noopener nofollow">this https URL</a>

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: currently, autonomous driving systems are insufficient in understanding the driving intentions of other vehicles, especially in recognizing taillight signals. Accurately recognizing taillight signals (such as brake lights and turn signals) is crucial for predicting vehicle behavior and preventing collisions. However, the existing open - source taillight datasets are scarce, small - scale, and inconsistently labeled, making it difficult to train and evaluate deep - learning models for reliable taillight detection. Specifically, the paper aims to: 1. **Fill the data set gap**: Provide a large - scale, diverse taillight data set (TLD), covering different traffic scenarios, lighting conditions, weather conditions, etc., to support the training and evaluation of taillight detection models. 2. **Improve taillight recognition accuracy**: By introducing a new two - stage vehicle taillight detection model, including a vehicle detection module and a taillight classifier, to more accurately identify the states of brake lights and turn signals. 3. **Enhance the safety of autonomous driving systems**: Enable autonomous driving systems to better understand the driving intentions of surrounding vehicles, thereby making safer and more intelligent decisions. ### Main problem summary - **Lack of high - quality data sets**: The existing taillight data sets are small - scale and inconsistently labeled, unable to meet the requirements of deep - learning model training. - **Complex taillight recognition challenges**: Taillight recognition faces multiple challenges, such as changes in lighting conditions, occlusions, and differences in taillight shapes of different vehicle models. - **Real - time and accuracy requirements**: Autonomous driving systems need to achieve real - time and accurate taillight state recognition in complex environments. By solving these problems, the paper hopes to provide strong support for the development of autonomous driving technology, especially in enhancing the safety and reliability of interactions between vehicles.

TLD: A Vehicle Tail Light signal Dataset and Benchmark

VLS: Vehicle Tail Light Signal Detection Benchmark

A Highly Efficient Vehicle Taillight Detection Approach Based on Deep Learning

A Robust Video Based Traffic Light Detection Algorithm for Intelligent Vehicles

TLD-READY: Traffic Light Detection -- Relevance Estimation and Deployment Analysis

Recognition of Front-Vehicle Taillights Based on YOLOv5s

A Vision-based Taillight Intention Detection Method for Intelligent and Connected Vehicles

Patterns of Vehicle Lights: Addressing Complexities in Curation and Annotation of Camera-Based Vehicle Light Datasets and Metrics

Enhanced Vision-Based Taillight Signal Recognition for Analyzing Forward Vehicle Behavior

Patterns of vehicle lights: Addressing complexities of camera-based vehicle light datasets and metrics

An Attention-based Recurrent Convolutional Network for Vehicle Taillight Recognition

Performance Evaluation of Region-Based Convolutional Neural Networks Toward Improved Vehicle Taillight Detection

Nighttime Driver Behavior Prediction Using Taillight Signal Recognition via CNN-SVM Classifier

An Indian Roads Dataset for Supported and Suspended Traffic Lights Detection

An Improved Approach for Real-Time Taillight Intention Detection by Intelligent Vehicles

VATLD: A Visual Analytics System to Assess, Understand and Improve Traffic Light Detection

Preceding vehicle detection and tracking adaptive to illumination variation in night traffic scenes based on relevance analysis

TLDM: An Enhanced Traffic Light Detection Model Based on YOLOv5

Real-time Detection of Vehicle and Traffic Light for Intelligent and Connected Vehicles Based on YOLOv3 Network

Provident Vehicle Detection at Night: The PVDN Dataset

DualCam: A Novel Benchmark Dataset for Fine-grained Real-time Traffic Light Detection