Deep Multi-Branch Aggregation Network for Real-Time Semantic Segmentation in Street Scenes

Weng, Xi; Yan, Yan; Dong, Genshun; Chen, Shu; Wang, Hanzi; Ji, Zhang

doi:10.1109/tits.2022.3150350

Cited by 26 publications

(8 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…But it increased computational load and consequently reduced inference speed. DMANet [35] integrated a feature fusion module into the overall structure to preserve the integrity of inactive spatial features. ViT [36] introduced a position embedding method to guide the segmentation task of input 2D image patches.…”

Section: Real-time Semantic Segmentationmentioning

confidence: 99%

LACTNet: A Lightweight Real-time Semantic Segmentation Network Based on Aggregation CNN and Transformer

Zhang,

Li,

et al. 2024

Preprint

View full text Add to dashboard Cite

Transformers have shown a significant advantage over CNNs in modeling long-range dependencies, which has led to their increased attention in semantic segmentation tasks. In the current work, a novel semantic segmentation model-LACTNet is introduced, which synergistically combines Transformer and CNN architectures for real-time processing. LACTNet is designed with a lightweight Transformer featuring a gated convolutional feedforward network, which is combined with CNNs to compensate for their respective shortcomings. LACTNet designs a Lightweight Average Feature Bottleneck (LAFB) module that effectively guides spatial detail information within the features, thus enhancing segmentation accuracy. To address the loss of spatial features in the decoder, a long skip-connection approach is employed through the designed Feature Fusion Enhancement Module (FFEM), which boosts both the integrity of spatial features and the feature interaction capability in the decoder. Testing on both the Cityscapes and CamVid datasets confirms that LACTNet attains mIoU scores of 74.8% and 71.8%, respectively, while also maintaining real-time frame rates of 90 FPS and 126 FPS.

show abstract

Section: Real-time Semantic Segmentationmentioning

confidence: 99%

LACTNet: A Lightweight Real-time Semantic Segmentation Network Based on Aggregation CNN and Transformer

Zhang,

Li,

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…STDC [20] formulates convolutional units that are better suited for semantic segmentation missions and introduces an auxiliary loss function to enable the network to concentrate on grasping spatial data. DMANet [21] develops a decoder for multi-branch aggregation alongside a lattice enhanced residual block to improve the network's feature representation. Atrous convolution assists in enlarging the sensory field and preserving additional spatial information.…”

Section: B Real-time Semantic Segmentationmentioning

confidence: 99%

Deep Multi - Resolution Network for Real- Time Semantic Segmentation in Street Scenes

Wang

Chen

Bian

et al. 2023

2023 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

Semantic segmentation is an essential technology for self-driving cars to comprehend their surroundings. Currently, real-time semantic segmentation networks commonly employ either encoder-decoder architecture or two-pathway architecture. Generally speaking, encoder-decoder models tend to be quicker, whereas two-pathway models exhibit higher accuracy. To leverage both strengths, we present the Spatial-Assistant Encoder-Decoder Network (SANet) to fuse the two architectures. In the overall architecture, we uphold the encoder-decoder design while maintaining the feature maps in the middle section of the encoder and utilizing atrous convolution branches for same-resolution feature extraction. Toward the end of the encoder, we integrate the asymmetric pooling pyramid pooling module (APPPM) to optimize the semantic extraction of the feature maps. This module incorporates asymmetric pooling layers that extract features at multiple resolutions. In the decoder, we present a hybrid attention module, SAD, that integrates horizontal and vertical attention to facilitate the combination of various branches. To ascertain the effectiveness of our approach, our SANet model achieved competitive results on the real-time CamVid and cityscape datasets. By employing a single 2080Ti GPU, SANet achieved a 78.4 % mIOU at 65.1 FPS on the Cityscape test dataset and 78.8 % mIOU at 147 FPS on the CamVid test dataset. The training code and model for SANet are available at https://github.com/CuZaoo/SANet-main

show abstract

“…Liu et al proposed the CRNet [12] segmentation network with a crossreference mechanism that enhances model output feature representations by comparing similar features in two images, achieving small-sample image segmentation. Weng et al introduced the DMA-Net [13] semantic segmentation network suitable for street view data, which aggregates feature maps generated by different convolutional layers through a multibranch aggregation network to obtain multi-scale information of the target, achieving good performance on the CamVid dataset.…”

Section: B Scene Segmentationmentioning

confidence: 99%

Construction Site Fence Recognition Method Based on Multi-Scale Attention Fusion ENet Segmentation Network (S)

Zhang,

Tang,

et al. 2023

International Conferences on Software Engineering and Knowledge Engineering

View full text Add to dashboard Cite

In this paper, we propose a fence recognition method based on the ENet (Efficient neural Network) segmentation network to address the problems of traditional segmentation networks, which have poor performance in recognizing fences with a large range of scale variations and hollow structures. Firstly, a multi-scale attention fusion ENet segmentation network is designed, which is trained using the fence with obvious color features. Then, a morphological algorithm is used to process the predicted image to restore the fence segmentation results. The designed multi-scale attention fusion segmentation network performs better on fence datasets than traditional methods. In addition, the activation function Leaky_Relu6 further enhances the stability and generalization ability of the network. The experiments are conducted on 540 fence images from different construction sites, and the computed IoU is 90%. The processing speed is about 28 frames per second. The experimental results show that our proposed network outperforms traditional segmentation algorithms in fence recognition performance, and achieves robustness in different construction scenarios while meeting the requirements of both accuracy and speed.

show abstract

Deep Multi-Branch Aggregation Network for Real-Time Semantic Segmentation in Street Scenes

Cited by 26 publications

References 39 publications

LACTNet: A Lightweight Real-time Semantic Segmentation Network Based on Aggregation CNN and Transformer

LACTNet: A Lightweight Real-time Semantic Segmentation Network Based on Aggregation CNN and Transformer

Deep Multi - Resolution Network for Real- Time Semantic Segmentation in Street Scenes

Construction Site Fence Recognition Method Based on Multi-Scale Attention Fusion ENet Segmentation Network (S)

Contact Info

Product

Resources

About