Robust Lane Detection via Expanded Self Attention

Lee, Minhyeok; Lee, Junhyeop; Lee, Dogyoon; Kim, Woojin; Hwang, Sangwon; Lee, Sangyoun

doi:10.48550/arxiv.2102.07037

Cited by 4 publications

(6 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Rich contextual information for further representation is encoded from a reasonable level of attention map. Expanded self attention (ESA) [24] designed for segmentation based lane detection in occluded and low-light images. The ESA module predicts the confidence of the lane by extracting global contextual information.…”

Section: Lane Line Segmentationmentioning

confidence: 99%

Joint Semantic Understanding with a Multilevel Branch for Driving Perception

Lee

Kim

2022

Applied Sciences

View full text Add to dashboard Cite

Visual perception is a critical task for autonomous driving. Understanding the driving environment in real time can assist a vehicle in driving safely. In this study, we proposed a multi-task learning framework for simultaneous traffic object detection, drivable area segmentation, and lane line segmentation in an efficient way. Our network encoder extracts features from an input image and three decoders at multilevel branches handle specific tasks. The decoders share the feature maps with more similar tasks for joint semantic understanding. Multiple loss functions are automatically weighted summed to learn multiple objectives simultaneously. We demonstrate the effectiveness of this framework on a BerkeleyDeepDrive100K( BDD100K) dataset. In the experiment, the proposed method outperforms the competing multi-task and single-task methods in terms of accuracy and maintains a real-time inference at more than 37 frames per second.

show abstract

Section: Lane Line Segmentationmentioning

confidence: 99%

Joint Semantic Understanding with a Multilevel Branch for Driving Perception

Lee

Kim

2022

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…Attention-based methods. Several attention-based lane detection models (Lee et al 2021;Tabelini et al 2020b;Liu et al 2021) have been proposed to capture the long-range information. (Lee et al 2021) propose a self-attention mechanism to predict the lanes' confidence along with the vertical and horizontal directions in an image.…”

Section: Related Workmentioning

confidence: 99%

“…The most state-of-the-art lane detection methods (Lee et al 2021;Xu et al 2020;Chen, Liu, and Lian 2019) Figure 1: Sketch map of proposed detection attention and row-column attention in Laneformer. Given the detected person and vehicle instances, detection attention is performed to capture the implicit relationship between them and lanes, e.g., lanes are more likely to appear next to cars.…”

Section: Introductionmentioning

confidence: 99%

“…In addition, the fixed receptive field of CNN architecture limits the ability to incorporate relations for longrange lane points, making them hard to capture well the characteristics of lanes since the shapes are conceptually long and thin. Several attention-based lane detection models (Lee et al 2021;Tabelini et al 2020b;Liu et al 2021) have been also proposed to capture the long-range information. Nevertheless, the fixed attention routines can not adap-…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Laneformer: Object-Aware Row-Column Transformers for Lane Detection

Han

Deng

Cai

et al. 2022

AAAI

View full text Add to dashboard Cite

We present Laneformer, a conceptually simple yet powerful transformer-based architecture tailored for lane detection that is a long-standing research topic for visual perception in autonomous driving. The dominant paradigms rely on purely CNN-based architectures which often fail in incorporating relations of long-range lane points and global contexts induced by surrounding objects (e.g., pedestrians, vehicles). Inspired by recent advances of the transformer encoder-decoder architecture in various vision tasks, we move forwards to design a new end-to-end Laneformer architecture that revolutionizes the conventional transformers into better capturing the shape and semantic characteristics of lanes, with minimal overhead in latency. First, coupling with deformable pixel-wise self-attention in the encoder, Laneformer presents two new row and column self-attention operations to efficiently mine point context along with the lane shapes. Second, motivated by the appearing objects would affect the decision of predicting lane segments, Laneformer further includes the detected object instances as extra inputs of multi-head attention blocks in the encoder and decoder to facilitate the lane point detection by sensing semantic contexts. Specifically, the bounding box locations of objects are added into Key module to provide interaction with each pixel and query while the ROI-aligned features are inserted into Value module. Extensive experiments demonstrate our Laneformer achieves state-of-the-art performances on CULane benchmark, in terms of 77.1% F1 score. We hope our simple and effective Laneformer will serve as a strong baseline for future research in self-attention models for lane detection.

show abstract

Section: Introductionmentioning

confidence: 99%

Laneformer: Object-aware Row-Column Transformers for Lane Detection

Han¹,

Deng²,

Cai³

et al. 2022

Preprint

View full text Add to dashboard Cite

We present Laneformer, a conceptually simple yet powerful transformer-based architecture tailored for lane detection that is a long-standing research topic for visual perception in autonomous driving. The dominant paradigms rely on purely CNN-based architectures which often fail in incorporating relations of long-range lane points and global contexts induced by surrounding objects (e.g., pedestrians, vehicles). Inspired by recent advances of the transformer encoder-decoder architecture in various vision tasks, we move forwards to design a new end-to-end Laneformer architecture that revolutionizes the conventional transformers into better capturing the shape and semantic characteristics of lanes, with minimal overhead in latency. First, coupling with deformable pixel-wise selfattention in the encoder, Laneformer presents two new row and column self-attention operations to efficiently mine point context along with the lane shapes. Second, motivated by the appearing objects would affect the decision of predicting lane segments, Laneformer further includes the detected object instances as extra inputs of multi-head attention blocks in the encoder and decoder to facilitate the lane point detection by sensing semantic contexts. Specifically, the bounding box locations of objects are added into Key module to provide interaction with each pixel and query while the ROI-aligned features are inserted into Value module. Extensive experiments demonstrate our Laneformer achieves state-of-the-art performances on CULane benchmark, in terms of 77.1% F1 score. We hope our simple and effective Laneformer will serve as a strong baseline for future research in self-attention models for lane detection.

show abstract

Robust Lane Detection via Expanded Self Attention

Cited by 4 publications

References 0 publications

Joint Semantic Understanding with a Multilevel Branch for Driving Perception

Joint Semantic Understanding with a Multilevel Branch for Driving Perception

Laneformer: Object-Aware Row-Column Transformers for Lane Detection

Laneformer: Object-aware Row-Column Transformers for Lane Detection

Contact Info

Product

Resources

About