VoxelEmbed: 3D Instance Segmentation and Tracking with Voxel Embedding based Deep Learning

Zhao, Mengyang; Liu, Quan; Jha, Aadarsh; Deng, Ruining; Yao, Tianyuan; Mahadevan‐Jansen, Anita; Tyska, Matthew J.; Millis, Bryan A.; Huo, Yuankai

doi:10.48550/arxiv.2106.11480

Cited by 3 publications

(2 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Image processing based on deep learning has been applied invarious complex industrial situations [29]. A simple multistream learning method is proposed in [30], which implements an end-to-end framework for one-stage 3D cell instance segmentation and tracking without numerous parameter adjustments. The algorithm adopts novel seed selection and early stopping methods to deal with the Euclidean distance metric, which greatly improves the computing speed and reduces the graphics processing unit (GPU) memory consumption [31].…”

Section: Related Workmentioning

confidence: 99%

A fine extraction algorithm for image-based surface cracks in underwater dams

Xin¹,

Fan²,

Shi³

et al. 2022

Meas. Sci. Technol.

View full text Add to dashboard Cite

Under the action of water erosion and self-aging action, reservoir dams are prone to have cracks, affecting the safe operation. Underwater visual imaging can be used to detect dam surface cracks, but the spalling, aquatic plants and suspended sediments result in low image contrast and complex backgrounds. With the use of unsupervised machine learning, this paper proposes a fine segmentation and extraction algorithm for image-based dam surface cracks. Firstly, the adaptive histogram equalization is used to make the uneven illumination areas of underwater surface images be even illumination areas, whose statistical characteristics is calculated under the linear spatial filtering. Secondly, the extraction problem of interested crack areas after dodging preprocess is transformed into calculating the distance of image block cluster center, which can distinguish the image blocks of crack features from the background interference features. Thirdly, the fine extraction of crack images is carried out in consideration of the connected domains and morphological features, and the posterior probability of image sample category is obtained based on the soft clustering of Gaussian mixed model. Finally, different extraction algorithms related to surface cracks are evaluated in extensive experiments. The results validate the superior performance of the proposed extraction algorithm with 90.1% extraction accuracy, 6.5% missing alarm rate, and 7.2% false alarm rate.

show abstract

Section: Related Workmentioning

confidence: 99%

A fine extraction algorithm for image-based surface cracks in underwater dams

Xin¹,

Fan²,

Shi³

et al. 2022

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…With the development of high-performance GPU, deep learning techniques have been widely applied to industrial inspection fields including steel [11][12][13][14][15][16][17][18][19], tire [20][21][22][23] and fabric [24,25] defect detection and other fields [26][27][28][29]. These methods can be divided into image classification [11][12][13], object detection [14][15][16] and segmentation methods [17][18][19].…”

Section: Introductionmentioning

confidence: 99%

CASPPNet: a chained atrous spatial pyramid pooling network for steel defect detection

Zheng¹,

Hu²,

Zhang³

et al. 2022

Meas. Sci. Technol.

View full text Add to dashboard Cite

Automatic machine vision-based defect detection has been successfully applied to many industrial visual inspection applications. However, automatic steel surface defect detection is still a challenging task due to diverse defect categories, low-contrast between defect and complex texture background. To address these challenges, a chained atrous spatial pyramid pooling network (CASPPNet) is proposed for steel surface defect detection. In CASPPNet, chained atrous spatial pyramid pooling is designed to enlarge receptive field and obtain enrich semantic information. An improved global attention feature fusion module is introduced to achieve feature interaction and salience. Moreover, residual boundary refinement block is introduced to get more complete defect boundary. Comparative experimental results verify that our method is superior to the state-of-the-art segmentation methods on public accessible SD-saliency-900 datasets and can meet the requirement of real-time online detection (the detection efficiency is at over 47 FPS on a single GPU).

show abstract

Multi-view fusion-based local-global dynamic pyramid convolutional cross-tansformer network for density classification in mammography

Zhong,

Piao,

Zhang

2023

Phys. Med. Biol.

View full text Add to dashboard Cite

Object: Breast density is an important indicator of breast cancer risk. However, existing methods for breast density classification do not fully utilise the multi-view information produced by mammography and thus have limited classification accuracy. Method: In this paper, we propose a multi-view fusion network, denoted local-global dynamic pyramidal-convolution transformer network (LG-DPTNet), for breast density classification in mammography. First, for single-view feature extraction, we develop a dynamic pyramid convolutional network to enable the network to adaptively learn global and local features. Second, we address the problem exhibited by traditional multi-view fusion methods, this is based on a cross-transformer that integrates fine-grained information and global contextual information from different views and thereby provides accurate predictions for the network. Finally, we use an asymmetric focal loss function instead of traditional cross-entropy loss during network training to solve the problem of class imbalance in public datasets, thereby further improving the performance of the model. Results: We evaluated the effectiveness of our method on two publicly available mammography datasets, CBIS-DDSM and INbreast, and achieved areas under the curve (AUC) of 96.73% and 91.12%, respectively. Conclusion: Our experiments demonstrated that the devised fusion model can more effectively utilise the information contained in multiple views than existing models and exhibits classification performance that is superior to that of baseline and state-of-the-art methods.

show abstract

VoxelEmbed: 3D Instance Segmentation and Tracking with Voxel Embedding based Deep Learning

Cited by 3 publications

References 25 publications

A fine extraction algorithm for image-based surface cracks in underwater dams

A fine extraction algorithm for image-based surface cracks in underwater dams

CASPPNet: a chained atrous spatial pyramid pooling network for steel defect detection

Multi-view fusion-based local-global dynamic pyramid convolutional cross-tansformer network for density classification in mammography

Contact Info

Product

Resources

About