Worsening Perception: Real-Time Degradation of Autonomous Vehicle
                    Perception Performance for Simulation of Adverse Weather
                    Conditions

Fursa, Ivan; Elias, Fandi,; Musat, Valentina; Culley, Jacob; Esteller, Enric Gil; Louise, Bilous,; Sluis, Isaac Vander; Rast, Alexander; Bradley, Andrew P.

doi:10.4271/12-05-01-0008

Cited by 5 publications

(7 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Despite being the most accurate and most commonly used class of detectors, however, CNN-based models display a significantly degraded performance in adverse weather conditions [13]. While this can be mitigated by training the model on datasets that contain adverse weather, the collecting and labelling of such datasets is expensive, laborious, and at the mercy of the prevailing weather conditions.…”

Section: Object Detection In Autonomous Racingmentioning

confidence: 99%

“…Numerous approaches were proposed to generate synthetic adverse weather images. Simple approaches have utilised open-source frameworks like OpenGL [34], and image augmentation libraries including [13], however, they lack visual realism. More sophisticated methods were proposed to generate more realistic adverse weather images.…”

Section: Synthetic Weather Generationmentioning

confidence: 99%

“…Although such models might perform well on simulation images, they may fail on images coming from real-world unconstrained environments. [13] who employed existing simple open-source augmentation libraries to generate adverse weather datasets, but likewise potentially suffers from similar issues. Other methods including [33,46] used GANs to improve tasks downstream from perception in urban and rural autonomous driving.…”

Section: Adverse Weather In Autonomous Racingmentioning

confidence: 99%

“…In the particular application of AV racing, a small error in perception may result in a considerable deviation from the optimal path [21] and a loss of time or even penalties or accidents. In both on-road and racing applications, the AV industry has a critical and urgent need to mitigate or deal with the significant degradation of perception caused by adverse weather conditions [39,13].…”

Section: Introductionmentioning

confidence: 99%

“…It also introduces a very costly and laborious process of (manually) labelling the collected data for different tasks like detection or segmentation. To overcome this challenge, researchers have suggested synthesising adverse weather datasets using simulators [52] or simple augmentation algorithms [13]. However, the applicability of these approaches for enhancement of object detection in a real-world setting is questionable, due to the lack of visual realism in the imagery.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Vision in adverse weather: Augmentation using CycleGANs with various object detectors for robust perception in autonomous racing

Teeti¹,

Musat²,

Khan³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

In an autonomous driving system, perception -identification of features and objects from the environment -is crucial. In autonomous racing, high speeds and small margins demand rapid and accurate detection systems. During the race, the weather can change abruptly, causing significant degradation in perception, resulting in ineffective manoeuvres. In order to improve detection in adverse weather, deep-learning-based models typically require extensive datasets captured in such conditions -the collection of which is a tedious, laborious, and costly process. However, recent developments in CycleGAN architectures allow the synthesis of highly realistic scenes in multiple weather conditions. To this end, we introduce an approach of using synthesised adverse condition datasets in autonomous racing (generated using CycleGAN) to improve the performance of four out of five state-of-the-art detectors by an average of 42.7 and 4.4 mAP percentage points in the presence of night-time conditions and droplets, respectively. Furthermore, we present a comparative analysis of five object detectors -identifying the optimal pairing of detector and training data for use during autonomous racing in challenging conditions.

show abstract

Section: Object Detection In Autonomous Racingmentioning

confidence: 99%

Section: Synthetic Weather Generationmentioning

confidence: 99%

Section: Adverse Weather In Autonomous Racingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Vision in adverse weather: Augmentation using CycleGANs with various object detectors for robust perception in autonomous racing

Teeti¹,

Musat²,

Khan³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Varied Realistic Autonomous Vehicle Collision Scenario Generation

Priisalu¹,

Păduraru²,

Smichisescu³

2023

Image Analysis

View full text Add to dashboard Cite

Object Detection in Adverse Weather for Autonomous Driving Through Data Merging and YOLOv8

Kumar,

Muhammad

2023

Preprint

View full text Add to dashboard Cite

For autonomous driving, perception is a primary and essential element that fundamentally deals with the insight into the ego vehicle’s environment through sensors. Perception is a challenging task suffering from dynamic objects and continuous environmental changes. The issue gets worse due to interrupting the quality of perception by adverse weather like snow, rain, fog, night light, sand storm, strong daylight, etc. In this work, we have tried to improve camera-based perception accuracy, such as autonomous driving-related object detection in adverse weather. We proposed the improvement of YOLOv8-based object detection in adverse weather through transfer learning using merged data from various harsh weather datasets. Two prosperous open-source datasets (ACDC and DAWN) and their merged dataset were used to detect primary objects on the road in harsh weather. A set of training weights were collected from training on the individual datasets, their merged version, and several subsets of those datasets according to their characteristics. A comparison between the training weights also occurred by evaluating the detection performance on the above-mentioned datasets and their subsets. The evaluation revealed that using custom datasets for training significantly improves the detection performance compared to the YOLOv8 base weights. And using more images through the feature-related data merging technique steadily increases the object detection performance.

show abstract

Worsening Perception: Real-Time Degradation of Autonomous Vehicle Perception Performance for Simulation of Adverse Weather Conditions

Cited by 5 publications

References 15 publications

Vision in adverse weather: Augmentation using CycleGANs with various object detectors for robust perception in autonomous racing

Vision in adverse weather: Augmentation using CycleGANs with various object detectors for robust perception in autonomous racing

Varied Realistic Autonomous Vehicle Collision Scenario Generation

Object Detection in Adverse Weather for Autonomous Driving Through Data Merging and YOLOv8

Contact Info

Product

Resources

About