Key Points Estimation and Point Instance Segmentation Approach for Lane Detection

  title={Key Points Estimation and Point Instance Segmentation Approach for Lane Detection},
  author={Yeongmin Ko and Younkwan Lee and Shoaib Azam and Farzeen Munir and Moongu Jeon and Witold Pedrycz},
  journal={IEEE Transactions on Intelligent Transportation Systems},
Perception techniques for autonomous driving should be adaptive to various environments. In essential perception modules for traffic line detection, many conditions should be considered, such as a number of traffic lines and computing power of the target system. To address these problems, in this paper, we propose a traffic line detection method called Point Instance Network (PINet); the method is based on the key points estimation and instance segmentation approach. The PINet includes several… 

Figures and Tables from this paper

Lane Detection Model Based on Spatio-Temporal Network with Double ConvGRUs

A spatio-temporal network with double Convolutional Gated Recurrent Units (ConvGRUs) is proposed to address lane detection in challenging scenes and can outperform the state-of-the-art lane detection models.

Real-Time Lane Instance Segmentation Using SegNet and Image Processing

This approach generates binary segmentation of lanes, then thebinary segmentation map is further processed to separate lanes, and a sliding window extracts each lane to produce the lane instance segmentation image.

RONELD: Robust Neural Network Output Enhancement for Active Lane Detection

This paper presents a real-time robust neural network output enhancement for active lane detection (RONELD) method to identify, track, and optimize active lanes from deep learning probability map outputs and hypothesizes true active lanes through tracking preceding frames.

Lane detection in complex scenes based on end-to-end neural network

This work proposes an end-to-end network to lane detection in a variety of complex scenes, and designs the information exchange block by combining spatial convolution and dilated convolution, which plays a great role in understanding detailed information.

Domain Adaptation For Lane Marking: An Unsupervised Approach

A lane marking model is designed that successfully generalizes from highway-traffic images to images of traffic data from Lahore (provincial capital of Pakistan) and shows that domain adaptation significantly improves the performance of lane marking system for the unlabelled data of the target domain.

End-to-end Lane Shape Prediction with Transformers

An end-to-end method that directly outputs parameters of a lane shape model, using a network built with a transformer to learn richer structures and context, showing its powerful deployment potential in real applications.

A Robust Lane Detection Model Using Vertical Spatial Features and Contextual Driving Information

Experimental results show that the proposed model can detect lane lines more robustly and precisely than state-of-the-art models in a variety of complex driving scenarios.

Training a thin and shallow lane detection network with self-knowledge distillation

Experiments prove that a small student network and large teacher network have a similar detection accuracy via SKD, and the student network has a shorter inference time and lower memory usage, and it can be applied for resource-limited devices flexibly.

LaneAF: Robust Multi-Lane Detection With Affinity Fields

This study presents an approach to lane detection involving the prediction of binary segmentation masks and per-pixel affinity fields, which can then be used to cluster lane pixels horizontally and vertically into corresponding lane instances in a post-processing step.



Towards End-to-End Lane Detection: an Instance Segmentation Approach

A fast lane detection algorithm, running at 50 fps, is proposed, which can handle a variable number of lanes and cope with lane changes, and is robust against road plane changes, unlike existing approaches that rely on a fixed, predefined transformation.

PointLaneNet: Efficient end-to-end CNNs for Accurate Real-Time Lane Detection

This paper proposes an end-to-end lane detection algorithm that can obtain the coordinates of lane line points directly from the drive scene, and can be deployed on the embedded platform NVIDIA PX2.

Improved Lane Detection With Multilevel Features in Branch Convolutional Neural Networks

The proposed lane detection network inspired by LaneNet model, which uses semantic segmentation concepts utilizes multiple level features of the encoder and designs enhanced binary segmentation and reduced pixel embedding branches achieves better accuracy performance than LaneNet and faster computation than the existed methods for real-time applications.

Multi-Class Lane Semantic Segmentation using Efficient Convolutional Networks

Two techniques, Feature Size Selection (FSS) and Degressive Dilation Block (DD Block) are proposed, which allow a network to extract thin lane features using appropriate feature sizes and acquire fine-grained spatial information.

Robust Lane Detection From Continuous Driving Scenes Using Deep Neural Networks

This work investigates lane detection by using multiple frames of a continuous driving scene, and proposes a hybrid deep architecture by combining the convolutional neural network and the recurrent neural network, which outperforms the competing methods in lane detection.

Lane detection using color-based segmentation

A new method based on color information is proposed that is robust in the presence of noise, shadows, pavement, and obstacles such like cars, motorcycles and pedestrians conditions and this method is applicable in complex environment.

End-to-end Lane Detection through Differentiable Least-Squares Fitting

This work proposes an end-to-end method where the features are optimized for the true task of interest: the network implicitly learns to generate features that prevent instabilities during the model fitting step, as opposed to two-step pipelines that need to handle outliers with heuristics.

Combining Priors, Appearance, and Context for Road Detection

An algorithm to estimate road priors online using geographical information is proposed, providing relevant initial information about the road location, and contextual cues, including horizon lines, vanishing points, lane markings, 3-D scene layout, and road geometry are used.

A Novel Strategy for Road Lane Detection and Tracking Based on a Vehicle’s Forward Monocular Camera

This paper aims to present the readers a novel strategy for lane detection and tracking, which fits as a functional requirement to deploy DAS features like Lane Departure Warning and Lane Keeping Assist.