FDDWNet: A Lightweight Convolutional Neural Network for Real-Time Semantic Segmentation

Liu, Jia; Zhou, Quan; Qiang, Yong; Kong, Bin; Wu, Xiaofu; Zheng, Baoyu

doi:10.1109/icassp40776.2020.9053838

Cited by 34 publications

(6 citation statements)

References 28 publications

(85 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, as this technology can replace conventional equipment, its utilization is expected to be higher. 53 introduced lightweight network: FDDWNet (factorized dilated depth-wise network). The key feature of this network is that its size is only 0.8 M, and its running speed is 60 FPS with a single RTX 2080TI GPU when the input size is 1024 × 512.…”

Section: Discussionmentioning

confidence: 99%

Stereo-vision-based 3D concrete crack detection using adversarial learning with balanced ensemble discriminator networks

Shim

Kim

Cho

et al. 2022

Structural Health Monitoring

View full text Add to dashboard Cite

The functional performance of concrete structures degrades over time as a result of continuous loads, stress fatigue, and external environmental changes. Thus, periodic diagnoses and inspections are essential because such conditions can eventually lead to disaster. Hence, the detection of cracks in concrete is a key component of structural management. In recent years, deep-learning-based computer vision technologies have emerged as a promising trend and have been actively used for crack detection. Unfortunately, the performance of existing crack detection technologies decreases under environmental conditions that vary widely. To resolve this issue, we propose a new deep neural network that applies an optimal mixing ratio of training data to improve recognition performance alongside an adversarial learning-based balanced ensemble discriminator network. Furthermore, a method to reconstruct the 3-dimensional shape of cracks is proposed using a stereo-vision-based triangulation measurement technique that determines the size of detected cracks. Experimental results show that the proposed algorithm achieved a crack detection performance with a mean intersection-over-union of 84.53% and an F1 score of 82.91%. The proposed inspection technology for concrete structures is expected to be implemented in the future in connection with various automation techniques.

show abstract

Section: Discussionmentioning

confidence: 99%

Stereo-vision-based 3D concrete crack detection using adversarial learning with balanced ensemble discriminator networks

Shim

Kim

Cho

et al. 2022

Structural Health Monitoring

View full text Add to dashboard Cite

show abstract

“…At the end of this part, visual comparison results will be presented. The performance of our EARMNet is tested with several state-of-art works in this part on the Cityscapes datasets: SegNet [6], Enet [18], SQNet [28], ESPNet [19], CGNet [38], ContextNet [39], EDANet [40], Fast-SCNN [42], Fast-SCNN [42], BiseNet [1], ICNet [43], DABNet [36], LEDNet [10], FBSNet [2], DFANet [44], FDDWNet [45] and MSCFNet [21]. We can learn from Table 5 and Table 6, the comparison results show that our EARMNet achieves a good balance between prediction accuracy and efficiency.…”

Section: Performance Evaluation On Cityscapesmentioning

confidence: 99%

EARMNet:A Fast and Accurate Semantic Segmentation Network with Lightweight Efficient Asymmetric Residual Module

Shen

Zhai

Chen³

et al. 2023

Preprint

View full text Add to dashboard Cite

Recently years, semantic segmentation has had extensive usage in many scenarios. Most real-world applications both require high prediction accuracy and running efficiency. To this end, we present a novel lightweight semantic segmentation method to pursue a balance between prediction accuracy and running efficiency. We first design our Efficient Asymmetric Residual Module (EARM) to capture local and context features with a lower computation cost. Following that, we introduce a novel Image Decomposition Attention Mechanism (IDAM) to fuse the feature map from different stages. Based on EARM, we arrange two EARM blocks with gradually increasing dilation rates as the main part of our method. Besides, we show a Lightweight Mutual-Attention Decoder (LMAD) to recover the detailed information effectively. Extensive experiments on Cityscapes and Camvid datasets show that our model achieves 73.3\% and 71.7\% mIoU along with 150 and 131 FPS on the Cityscapes and Camvid datasets respectively

show abstract

“…Each network was trained using a dataset containing blurred images at a mixing ratio of 50%. Te segmentation network used four types of lightweight networks: ERFNet [47], CGNet [48], LedNet [49], and FDDWNet [50]. Performance evaluation of the trained model for each learning structure was performed by increasing the ratio of blurred images in the test dataset to clarify the dependence the performance on the quality of the input image.…”

Section: Proposed Segmentation Frameworkmentioning

confidence: 99%

Learning Structure for Concrete Crack Detection Using Robust Super-Resolution with Generative Adversarial Network

Kim

Shim

Kang

et al. 2023

Structural Control and Health Monitoring

View full text Add to dashboard Cite

To develop countermeasures against the inevitable aging of tunnels, accurate inspection is critical for ensuring stable tunnel services. Tunnels are large-scale infrastructures, whereas the cracks in such tunnels are small-scale objects, necessitating the development of methods capable of rapidly detecting cracks in a wide field of view. Therefore, this study proposes a new learning structure based on a segmentation network. This structure includes super-resolution and generative adversarial networks, which contribute to improved detection performance with robustness to input data. Furthermore, a method for the effective construction of training data for the application of the proposed structure is presented. Subsequently, the performance of the method over 1,606 crack images with randomly degraded qualities is evaluated. The proposed structure presents improved crack intersection over union and F1-scores of 63.686% and 77.811%, respectively, and low variances of 0.9008 and 0.5015 compared to the original structure. The results presented herein indicate the possible application of the proposed accurate condition inspection technology to tunnel maintenance in the future.

show abstract

FDDWNet: A Lightweight Convolutional Neural Network for Real-Time Semantic Segmentation

Cited by 34 publications

References 28 publications

Stereo-vision-based 3D concrete crack detection using adversarial learning with balanced ensemble discriminator networks

Stereo-vision-based 3D concrete crack detection using adversarial learning with balanced ensemble discriminator networks

EARMNet:A Fast and Accurate Semantic Segmentation Network with Lightweight Efficient Asymmetric Residual Module

Learning Structure for Concrete Crack Detection Using Robust Super-Resolution with Generative Adversarial Network

Contact Info

Product

Resources

About