ST-MAE: robust lane detection in continuous multi-frame driving scenes based on a deep hybrid network
نویسندگان
چکیده
Abstract Lane detection is one of the key techniques to realize advanced driving assistance and automatic driving. However, lane networks based on deep learning have significant shortcomings. The results are often unsatisfactory when there shadows, degraded markings, vehicle occlusion lanes. Therefore, a continuous multi-frame image sequence network proposed. Specifically, six-frame input into network, in which scene information each frame extracted by an encoder composed Swin Transformer blocks PredRNN. Continuous modeled as time-series ST-LSTM blocks, then, shape changes motion trajectory spatiotemporal effectively modeled. Finally, through decoder features obtained reconstructed complete task. Extensive experiments two large-scale datasets demonstrate that proposed method outperforms competing methods detection, especially handling difficult situations. Experiments carried out TuSimple dataset. show: for easy scenes, validation accuracy 97.46%, test 97.37%, precision 0.865. For complex 97.38%, 97.29%, 0.859. running time 4.4 ms. CULane show that, 97.03%, 96.84%, 0.837. 96.18%, 95.92%, 0.829. 6.5
منابع مشابه
A robust hybrid method for text detection in natural scenes by learning-based partial differential equations
Learning-based partial differential equations (PDEs), which combine fundamental differential invariants into a non-linear regressor, have been successfully applied to several computer vision tasks. In this paper, we present a robust hybrid method that uses learning-based PDEs for detecting texts from natural scene images. Our method consists of both top-down and bottom-up processing, which are ...
متن کاملA Reliable and Robust Lane Detection System based on the Parallel Use of Three Algorithms for Driving Safety Assistance
Road traffic incidents analysis has shown that a third of them occurs without any conflict which indicates problems with road following. In this paper a driving safety assistance system is introduced, whose aim is to prevent the driver drifting off or running off the road. The road following system is based on a frontal on-board monocular camera. In order to get a high degree of reliability and...
متن کاملReal-time camera based lane detection for autonomous driving
In this report a new designed real-time lane detection algorithm based on monocular vision and inertial measurement unit (IMU) is explained and evaluated. The acquired image from the camera is undergoing an inverse perspective mapping to create a bird-view perspective of the road. Local maxima in this image determine the lane-markers. Faulty local maxima are removed and constrained line fitting...
متن کاملA study on robust detection of pronunciation erroneous tendency based on deep neural network
Compared with scoring feedbacks, instructive feedbacks are more demanded by language learners using computer aided pronunciation training (CAPT) systems, which require detailed information about erroneous pronunciations along with phone errors. Pronunciation erroneous tendency (PET) defines a set of incorrect articulation configurations regarding main articulators and uttering manners for the p...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Complex & Intelligent Systems
سال: 2022
ISSN: ['2198-6053', '2199-4536']
DOI: https://doi.org/10.1007/s40747-022-00909-0