The DriveU Traffic Light Dataset: Introduction and Comparison with Existing Datasets

Fregin, Andreas; Müller, Julian; Krebel, Ulrich; Diermayer, Klaus

doi:10.1109/icra.2018.8460737

Cited by 45 publications

(28 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The training and evaluation of the YOLOv3 detector leveraged two public available datasets: DriveU Traffic Light Dataset (DTLD) [29] and LISA Traffic Light Dataset (LISA-TLD) [7].…”

Section: A Datasets For Traffic Light Detectionmentioning

confidence: 99%

Traffic Light Recognition Using Deep Learning and Prior Maps for Autonomous Cars

Possatti

Guidolini

Cardoso

et al. 2019

2019 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

Autonomous terrestrial vehicles must be capable of perceiving traffic lights and recognizing their current states to share the streets with human drivers. Most of the time, human drivers can easily identify the relevant traffic lights. To deal with this issue, a common solution for autonomous cars is to integrate recognition with prior maps. However, additional solution is required for the detection and recognition of the traffic light. Deep learning techniques have showed great performance and power of generalization including traffic related problems. Motivated by the advances in deep learning, some recent works leveraged some state-of-the-art deep detectors to locate (and further recognize) traffic lights from 2D camera images. However, none of them combine the power of the deep learning-based detectors with prior maps to recognize the state of the relevant traffic lights. Based on that, this work proposes to integrate the power of deep learning-based detection with the prior maps used by our car platform IARA (acronym for Intelligent Autonomous Robotic Automobile) to recognize the relevant traffic lights of predefined routes. The process is divided in two phases: an offline phase for map construction and traffic lights annotation; and an online phase for traffic light recognition and identification of the relevant ones. The proposed system was evaluated on five test cases (routes) in the city of Vitória, each case being composed of a video sequence and a prior map with the relevant traffic lights for the route. Results showed that the proposed technique is able to correctly identify the relevant traffic light along the trajectory.

show abstract

“…The training and evaluation of the YOLOv3 detector leveraged two public available datasets: DriveU Traffic Light Dataset (DTLD) [29] and LISA Traffic Light Dataset (LISA-TLD) [7].…”

Section: A Datasets For Traffic Light Detectionmentioning

confidence: 99%

Traffic Light Recognition Using Deep Learning and Prior Maps for Autonomous Cars

Possatti

Guidolini

Cardoso

et al. 2019

2019 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

show abstract

“…Images with a resolution of two megapixel and annotated traffic light metadata are used for our experiments from the DriveU dataset [33]. Furthermore, we have annotated the left camera images with additional metadata: lane arrow markings, lane signs, and lane line markings for all visible lanes, cf .…”

Section: A Databasementioning

confidence: 99%

Deep Metadata Fusion for Traffic Light to Lane Assignment

Langenberg

Lüddecke

Wörgötter

2019

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

We present a deep metadata fusion approach that connects image data and heterogeneous metadata inside a Convolutional Neural Network (CNN). This approach enables us to assign all relevant traffic lights to their associated lanes. To achieve this, a common CNN topology is trained by down-sampled and transformed input images to predict an indication vector. The indication vector contains the column positions of all the relevant traffic lights that are associated with lanes. In parallel, we fuse prepared and adaptively weighted Metadata Feature Maps (MFM) with the convolutional feature map input of a selected convolutional layer. The results are compared to rule-based, only-metadata, and only-vision approaches. In addition, human performance of the traffic light to ego-vehicle lane assignment has been measured by a subjective test. The proposed approach outperforms all other approaches. It achieves about 93.0% average precision for a real-world dataset. In a more complex dataset, 87.1% average precision is achieved. In particular, the new approach reaches significantly higher results with 93.7% to 91.0% average accuracy for a real-world dataset in contrast to lower human performance.

show abstract

“…The experiments were run on the DriveU Traffic Light Dataset (DTLD) [Fregin et al 2018], the largest publicly available dataset of traffic lights. DTLD was assembled based on daytime records of 11 German cities in different weather conditions.…”

Section: Training and Test Datasetsmentioning

confidence: 99%

“…More recently, state-of-the-art deep networks were used to jointly perform the detection and state classification of traffic lights [Behrendt et al 2017, Jensen et al 2017, Pon et al 2018. Despite the significant advances in solving TLR, when relying on images of the scene alone, little progress was observed for a more specific and interesting (considering real world applications) problem: the recognition of the state focusing on the relevant traffic lights [Fregin et al 2018], i.e., a subset of the visible traffic lights that define whether or not the vehicle can continue on its way.…”

Section: Introductionmentioning

confidence: 99%

Relevant Traffic Light Localization via Deep Regression

Freitas

Paixão

Berriel

et al. 2019

Anais Do XVI Encontro Nacional De Inteligência Artificial E Computacional (ENIAC 2019)

View full text Add to dashboard Cite

Artificial intelligence advances have an important role on self-driving cars development, such as assisting the recognition of traffic lights. However, when relying on images of the scene alone, little progress was observed on selecting the traffic lights defining guidance to the car. Common detection approaches rely on additional high-level decision-making process to select a relevant traffic light. This work address the problem by proposing a deep regression system with an outliers resilient loss to predict the coordinates of a relevant traffic light in the image plane. The prediction can be used as a high-level decision-maker or as an assistant to a cheaper classifier to work on a region of interest. Results for European scenes show success in about 88% of the cases.

show abstract

The DriveU Traffic Light Dataset: Introduction and Comparison with Existing Datasets

Cited by 45 publications

References 22 publications

Traffic Light Recognition Using Deep Learning and Prior Maps for Autonomous Cars

Traffic Light Recognition Using Deep Learning and Prior Maps for Autonomous Cars

Deep Metadata Fusion for Traffic Light to Lane Assignment

Relevant Traffic Light Localization via Deep Regression

Contact Info

Product

Resources

About