Lightweight and Efficient Convolutional Neural Network for Road Scene Semantic Segmentation

Kherraki, Amine; Maqbool, Muaz; Ouazzani, Rajae El

doi:10.1109/iccp56966.2022.10053967

Cited by 1 publication

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Section: Methodsmentioning

confidence: 99%

“…The four methods in Table 1 are: the original DDRNet-23 as the baseline and our proposed DDF&SE-DDRNet. In addition, Table 1 shows the segmentation effects of the other three state-of-the-art methods, LECNN [12], BiSeNet V2 [20], and ResNet-50 [21]. 1 shows the PA value, mPA value, MIou value, FPS value and the number of parameters in the network for the four segmentation methods on the Cityscapes dataset.…”

Section: Road Scene Segmentation On Cityscape Datasetmentioning

confidence: 99%

“…However, the experimental results of this network on the Cityscapes dataset are not ideal. Kherraki et al [ 12 ] believe that accuracy is not the first consideration in road scene segmentation tasks. They proposed a new network based on an encoder–decoder architecture, which reduces accuracy with fewer parameters.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Deep Dual-Resolution Road Scene Segmentation Networks Based on Decoupled Dynamic Filter and Squeeze–Excitation Module

Sun

2023

Sensors

View full text Add to dashboard Cite

Image semantic segmentation is an important part of automatic driving assistance technology. The complexity of road scenes and the real-time requirements of application scenes for segmentation algorithm are the challenges facing segmentation algorithms. In order to meet the above challenges, Deep Dual-resolution Road Scene Segmentation Networks based on Decoupled Dynamic Filter and Squeeze–Excitation (DDF&SE-DDRNet) are proposed in this paper. The proposed DDF&SE-DDRNet uses decoupled dynamic filter in each module to reduce the number of network parameters and enable the network to dynamically adjust the weight of each convolution kernel. We add the Squeeze-and-Excitation module to each module of DDF&SE-DDRNet so that the local feature map in the network can obtain global features to reduce the impact of image local interference on the segmentation result. The experimental results on the Cityscapes dataset show that the segmentation accuracy of DDF&SE-DDRNet is at least 2% higher than that of existing algorithms. Moreover, DDF&SE-DDRNet also has satisfactory inferring speed.

show abstract

Section: Methodsmentioning

confidence: 99%