Automatic labeling of vulnerable road users in multi-sensor data

Dimitrievski, Martin; Shopovska, Ivana; Hamme, David Van; Veelaert, Peter; Philips, Wilfried

doi:10.1109/itsc48978.2021.9564692

Cited by 6 publications

(10 citation statements)

References 17 publications

(22 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The colour images are annotated semi-automatically using an auto-labelling tool [28]. The annotations for thermal images are formed with the help of colour images' annotations with negligible parallax, using the stereo projection method [29,30].…”

Section: Methodsmentioning

confidence: 99%

Probabilistic Fusion for Pedestrian Detection from Thermal and Colour Images

Shaikh

Hamme

Veelaert

et al. 2022

Sensors

View full text Add to dashboard Cite

Pedestrian detection is an important research domain due to its relevance for autonomous and assisted driving, as well as its applications in security and industrial automation. Often, more than one type of sensor is used to cover a broader range of operating conditions than a single-sensor system would allow. However, it remains difficult to make pedestrian detection systems perform well in highly dynamic environments, often requiring extensive retraining of the algorithms for specific conditions to reach satisfactory accuracy, which, in turn, requires large, annotated datasets captured in these conditions. In this paper, we propose a probabilistic decision-level sensor fusion method based on naive Bayes to improve the efficiency of the system by combining the output of available pedestrian detectors for colour and thermal images without retraining. The results in this paper, obtained through long-term experiments, demonstrate the efficacy of our technique, its ability to work with non-registered images, and its adaptability to cope with situations when one of the sensors fails. The results also show that our proposed technique improves the overall accuracy of the system and could be very useful in several applications.

show abstract

Section: Methodsmentioning

confidence: 99%

Probabilistic Fusion for Pedestrian Detection from Thermal and Colour Images

Shaikh

Hamme

Veelaert

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…In the proposed method, all algorithm design choices consider the effect on the detection of traffic road users. For evaluation purposes, we relied on HDR images simulated from images in a public SDR traffic dataset [ 28 ], as well as our multi-modal traffic dataset including true HDR data with “person” annotations [ 29 ].…”

Section: Related Workmentioning

confidence: 99%

High-Dynamic-Range Tone Mapping in Intelligent Automotive Systems

Shopovska

Stojkovic

Aelterman

et al. 2023

Sensors

Self Cite

View full text Add to dashboard Cite

Intelligent driver assistance systems are becoming increasingly popular in modern passenger vehicles. A crucial component of intelligent vehicles is the ability to detect vulnerable road users (VRUs) for an early and safe response. However, standard imaging sensors perform poorly in conditions of strong illumination contrast, such as approaching a tunnel or at night, due to their dynamic range limitations. In this paper, we focus on the use of high-dynamic-range (HDR) imaging sensors in vehicle perception systems and the subsequent need for tone mapping of the acquired data into a standard 8-bit representation. To our knowledge, no previous studies have evaluated the impact of tone mapping on object detection performance. We investigate the potential for optimizing HDR tone mapping to achieve a natural image appearance while facilitating object detection of state-of-the-art detectors designed for standard dynamic range (SDR) images. Our proposed approach relies on a lightweight convolutional neural network (CNN) that tone maps HDR video frames into a standard 8-bit representation. We introduce a novel training approach called detection-informed tone mapping (DI-TM) and evaluate its performance with respect to its effectiveness and robustness in various scene conditions, as well as its performance relative to an existing state-of-the-art tone mapping method. The results show that the proposed DI-TM method achieves the best results in terms of detection performance metrics in challenging dynamic range conditions, while both methods perform well in typical, non-challenging conditions. In challenging conditions, our method improves the detection F2 score by 13%. Compared to SDR images, the increase in F2 score is 49%.

show abstract

“…A lot of effort nowadays is put into preparing Artificial Intelligence (AI) based auto-labeling algorithms, which can either replace or augment the manual labeling effort [21], [54]. This is important because even a small reduction in the time required to label one second of logged data results in huge savings in project time and money.…”

Section: Data Labeling (Annotation)mentioning

confidence: 99%

Intelligent Data Handling in Current and Next-Generation Automated Vehicle Development—A Review

Komorkiewicz¹,

Chin

Skruch

et al. 2023

IEEE Access

View full text Add to dashboard Cite

The higher the vehicle autonomy level is, the more the problems that need to be solved by engineers designing its sensors and control systems. Apart from the mechanical, electrical, algorithmic and validation challenges, such projects require to efficiently gather, store and handle the very large amounts of data logged from the vehicles. This is because the key functionalities of the control system in automated vehicles are based on data that is collected from a variety of sources, processed and analyzed to generate actuation signals. The article goes through the challenges, opportunities as well as solutions associated with data logging, collection, storage, annotation, reprocessing and evaluation. It highlights that effective and efficient data handling is an essential element for obtaining the required performance, reliability, safety and quality capabilities that would allow the mass production of automated vehicles. In terms of the safety of passengers and other road users, the behavior of a mass-produced automated vehicle must be predictable and more reliable than that of an ordinary driver. The multitude of variants that test sequences must cover is practically endless. Therefore, shortening the time-to-market to an acceptable length is only possible if you use the right methods of working with data.

show abstract

Automatic labeling of vulnerable road users in multi-sensor data

Cited by 6 publications

References 17 publications

Probabilistic Fusion for Pedestrian Detection from Thermal and Colour Images

Probabilistic Fusion for Pedestrian Detection from Thermal and Colour Images

High-Dynamic-Range Tone Mapping in Intelligent Automotive Systems

Intelligent Data Handling in Current and Next-Generation Automated Vehicle Development—A Review

Contact Info

Product

Resources

About