Keep your Eyes on the Lane: Real-time Attention-guided Lane Detection

Tabelini, Lucas; Berriel, Rodrigo F.; Paixão, Thiago M.; Badué, Claudine; Souza, Alberto F. De; Oliveira-Santos, Thiago

doi:10.1109/cvpr46437.2021.00036

Cited by 274 publications

(162 citation statements)

References 20 publications

Supporting

Mentioning

161

Contrasting

Unclassified

Order By: Relevance

“…Qu et al [26] estimated multiple keypoints and associated them to reconstruct actual lanes. Meanwhile, an anchor-based detection framework was employed for lane detection [18,30]. These anchor-based techniques consider straight lines as lane candidates (or anchors) and generate a predefined set of candidates.…”

Section: Related Workmentioning

confidence: 99%

“…SI computes the classification probability and regression offset of each candidate, while IC estimates the compatibility between each pair of lanes. Extensive experiments show that the proposed algorithm provides competitive results on existing datasets [1,22] and outperforms the state-of-the-art techniques [30,34] on a new dataset, called SDLane, containing structurally more diverse lanes.…”

Section: Introductionmentioning

confidence: 97%

“…However, even these algorithms may fail to detect less visible lanes in cluttered scenes, because they use only local features and may miss parts of lanes. Meanwhile, the anchor-based detection framework, which has been used extensively in object detection [15,20,27,28], has been recently adopted for lane detection [18,30]. These anchor-based algorithms consider straight lines as anchors (or lane candidates).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Eigenlanes: Data-Driven Lane Descriptors for Structurally Diverse Lanes

Jin¹,

Wonhui²,

Jeong³

et al. 2022

Preprint

View full text Add to dashboard Cite

A novel algorithm to detect road lanes in the eigenlane space is proposed in this paper. First, we introduce the notion of eigenlanes, which are data-driven descriptors for structurally diverse lanes, including curved, as well as straight, lanes. To obtain eigenlanes, we perform the best rank-M approximation of a lane matrix containing all lanes in a training set. Second, we generate a set of lane candidates by clustering the training lanes in the eigenlane space. Third, using the lane candidates, we determine an optimal set of lanes by developing an anchor-based detection network, called SIIC-Net. Experimental results demonstrate that the proposed algorithm provides excellent detection performance for structurally diverse lanes. Our codes are available at https://github.com/dongkwonjin/Eigenlanes.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 97%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Eigenlanes: Data-Driven Lane Descriptors for Structurally Diverse Lanes

Jin¹,

Wonhui²,

Jeong³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Meanwhile, in [16], lane markers were tracked temporally. Additionally, [21] presents an anchor-based single-stage deep lane detection model using anchors for feature pooling. In [17], the authors developed 3D-LaneNet, a network that predicts the 3D layout of lanes using a single image.…”

Section: Related Researchmentioning

confidence: 99%

LaneAF: Robust Multi-Lane Detection with Affinity Fields

Abualsaud¹,

Liu²,

Lu³

et al. 2021

Preprint

View full text Add to dashboard Cite

This study presents an approach to lane detection involving the prediction of binary segmentation masks and perpixel affinity fields. These affinity fields, along with the binary masks, can then be used to cluster lane pixels horizontally and vertically into corresponding lane instances in a post-processing step. This clustering is achieved through a simple row-by-row decoding process with little overhead; such an approach allows LaneAF to detect a variable number of lanes without assuming a fixed or maximum number of lanes. Moreover, this form of clustering is more interpretable in comparison to previous visual clustering approaches, and can be analyzed to identify and correct sources of error. Qualitative and quantitative results obtained on popular lane detection datasets demonstrate the model's ability to detect and cluster lanes effectively and robustly. Our proposed approach performs on par with stateof-the-art approaches on the limited TuSimple benchmark, and sets a new state-of-the-art on the challenging CULane dataset.

show abstract

“…However, lane detection also necessitates a realtime, lightweight solution that can be placed on edge devices in autonomous vehicles while being able to respond rapidly to changes in the dynamic and high-speed driving environment. While recent state-of-the-art lane detection methods have adopted new innovations, such as the attention mechanism [3] and in particular the self-attention mechanism [4], [5], to address some of these concerns, it is observed that many of these models still face difficulty adapting to datasets that differ significantly from their train sets. This covariate shift arising from the different distribution of train and test set data can be attributed to reasons such as varying road surface conditions and different lane markings.…”

Section: Introductionmentioning

confidence: 99%