检索结果-内蒙古大学图书馆

Dynamic Frame Interpolation in Wavelet Domain

IEEE Transactions on image processing 2023年 32卷 5296-5309页

作者： Kong, Lingtong Jiang, Boyuan Luo, Donghao Chu, Wenqing Tai, Ying Wang, Chengjie Yang, Jie Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Department of Automation Shanghai200240 China Tencent Youtu Laboratory Shanghai200233 China Nanjing University School of Intelligence Science and Technology Suzhou215163 China

Video frame interpolation is an important low-level vision task, which can increase frame rate for more fluent visual experience. Existing methods have achieved great success by employing advanced motion models and synthesis networks. However, the spatial redundancy when synthesizing the target frame has not been fully explored, that can result in lots of inefficient computation. On the other hand, the computation compression degree in frame interpolation is highly dependent on both texture distribution and scene motion, which demands to understand the spatial-temporal information of each input frame pair for a better compression degree selection. In this work, we propose a novel two-stage frame interpolation framework termed WaveletVFI to address above problems. It first estimates intermediate optical flow with a lightweight motion perception network, and then a wavelet synthesis network uses flow aligned context features to predict multi-scale wavelet coefficients with sparse convolution for efficient target frame reconstruction, where the sparse valid masks that control computation in each scale are determined by a crucial threshold ratio. Instead of setting a fixed value like previous methods, we find that embedding a classifier in the motion perception network to learn a dynamic threshold for each sample can achieve more computation reduction with almost no loss of accuracy. On the common high resolution and animation frame interpolation benchmarks, proposed WaveletVFI can reduce computation up to 40% while maintaining similar accuracy, making it perform more efficiently against other state-of-the-arts. © 1992-2012 IEEE.

关键词： Wavelet transforms

来源：评论

学校读者我要写书评

暂无评论

Average Weight Optimization RBPF Method for Target Tracking in Multi-Sensor Observations

引用

Chinese Journal of Electronics 2023年第2期22卷 401-404页

作者： Xianxing Liu Zhentao Hu Jie Li Institute of Image Processing and Pattern Recognition Henan University Kaifeng China

The reasonable design of particle filter framework in multi-sensor observation system is the key to expand the application domain of sampling nonlinear filters. Aiming at the effective realization of particle filter for multi-sensor target tracking problem, a novel average weight optimization Rao-Blackwellised particle filtering algorithm is proposed. Combining with the kinetic equation of target state evolution, RBPF is used as the basic estimator of algorithm realization. For the rational utilization from multi-sensor observations and the reduction of the adverse influence from random observations noise in measuring process of particles weight, the average weight optimization strategy is used to improve the reliability and stability of particle weight variance. In addition, we give the concrete flow of RBPF in average weight optimization strategy. Finally, the theoretical analysis and experimental results show the feasibility and efficiency of the proposed algorithm.

关键词： Weight measurement Technological innovation Target tracking Redundancy Filtering algorithms Reliability theory Particle filters

来源：评论

学校读者我要写书评

暂无评论

A Coarse-to-Fine Framework for Head and Neck Tumor Segmentation in CT and PET images 1

引用

2nd 3D Head and Neck Tumor Segmentation in PET/CT Challenge, HECKTOR 2021, held in conjunction with 24th International Conference on Medical image Computing and Computer-Assisted Intervention, MICCAI 2021

作者： An, Chengyang Chen, Huai Wang, Lisheng Department of Automation Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai200240 China

ISBN: (数字)9783030982539

ISBN: (纸本)9783030982522

Radiomics analysis can help patients suffered from head and neck (H&N) cancer customize tailoring treatments. It requires a large number of segmentation of the H&N tumor area in PET and CT images. However, the cost of manual segmentation is extremely high. In this paper, we propose a coarse-to-fine framework to segment the H&N tumor automatically in FluoroDeoxyGlucose (FDG)-Positron Emission Tomography (PET) and Computed Tomography (CT) images. Specifically, we trained three 3D-UNets with residual blocks to make coarse stage, fine stage and refined stage predictions respectively. Experiments show that such a training framework can improve the segmentation quality step by step. We evaluated our framework with Dice Similarity Coefficient (DSC) and Hausdorff Distance at 95% (HD95) of 0.7733 and 3.0882 respectively in the task 1 of the HEad and neCK TumOR segmentation and outcome prediction in PET/CT images (HECKTOR2021) Challenge and ranked second. © 2022, Springer Nature Switzerland AG.

关键词： Positron emission tomography

来源：评论

学校读者我要写书评

暂无评论

GAN-DIRNet:A Novel Deformable image Registration Approach for Multimodal Histological images

引用

Computers, Materials & Continua 2024年第7期80卷 487-506页

作者： Haiyue Li Jing Xie Jing Ke Ye Yuan Xiaoyong Pan Hongyi Xin Hongbin Shen Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong Universityand Key Laboratory of System Control and Information ProcessingMinistry of Education of ChinaShanghai200240China Department of Pathology Ruijin HospitalShanghai Jiao Tong UniversitySchool of MedicineShanghai200025China Department of Computer Science and Engineering Shanghai Jiao Tong UniversityShanghai200240China Global Institute of Future Technology Shanghai Jiao Tong UniversityShanghai200240China

Multi-modal histological image registration tasks pose significant challenges due to tissue staining operations causing partial loss and folding of *** neural network(CNN)and generative adversarial network(GAN)are pivotal inmedical image ***,existing methods often struggle with severe interference and deformation,as seen in histological images of conditions like Cushing’s *** argue that the failure of current approaches lies in underutilizing the feature extraction capability of the discriminator *** this study,we propose a novel multi-modal registration approach GAN-DIRNet based on GAN for deformable histological image *** begin with,the discriminators of two GANs are embedded as a new dual parallel feature extraction module into the unsupervised registration networks,characterized by implicitly extracting feature descriptors of specific ***,modal feature description layers and registration layers collaborate in unsupervised optimization,facilitating faster convergence and more precise ***,experiments and evaluations were conducted on the registration of the Mixed National institute of Standards and Technology database(MNIST),eight publicly available datasets of histological sections and the Clustering-Registration-Classification-Segmentation(CRCS)dataset on the Cushing’s *** results demonstrate that our proposed GAN-DIRNet method surpasses existing approaches like DIRNet in terms of both registration accuracy and time efficiency,while also exhibiting robustness across different image types.

关键词： Histological images registration deformable registration generative adversarial network cushing’s disease machine learning computer vision

来源：评论

学校读者我要写书评

暂无评论

Remaining-data-free Machine Unlearning by Suppressing Sample Contribution

arXiv

引用

arXiv 2024年

作者： Cheng, Xinwen Huang, Zhehao Zhou, Wenxin He, Zhengbao Yang, Ruikai Wu, Yingwen Huang, Xiaolin Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai200240 China

Machine unlearning (MU) is to forget data from a well-trained model, which is practically important due to the "right to be forgotten". The unlearned model should approach the retrained model, where the forgetting data are not involved in the training process and hence do not contribute to the retrained model. Considering the forgetting data’s absence during retraining, we think unlearning should withdraw their contribution from the pre-trained model. The challenge is that when tracing the learning process is impractical, how to quantify and detach sample’s contribution to the dynamic learning process using only the pre-trained model. We first theoretically discover that sample’s contribution during the process will reflect in the learned model’s sensitivity to it. We then practically design a novel method, namely MU-Mis (Machine Unlearning by Minimizing input sensitivity), to suppress the contribution of the forgetting data. Experimental results demonstrate that MU-Mis can unlearn effectively and efficiently without utilizing the remaining data. It is the first time that a remaining-data-free method can outperform state-of-the-art (SoTA) unlearning methods that utilize the remaining data. Copyright © 2024, The Authors. All rights reserved.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Based Neovascularization Detection in Color Fundus Photography image 23

Deep Learning Based Neovascularization Detection in Color Fu...

引用

9th International Conference on Computing and Artificial Intelligence, ICCAI 2023

作者： Guo, Tianjiao Yang, Jie Yu, Qi Institute of Medical Robotics Institute of Image Processing and Pattern Recognition Department of Automation School of Biomedical Engineering Shanghai Jiao Tong University Shanghai China Institute of Image Processing and Pattern Recognition Department of Automation Shanghai Jiao Tong University Shanghai China Department of Ophthalmology Shanghai General Hospital Shanghai Jiao Tong University School of Medicine Shanghai China National Clinical Research Center for Eye Diseases Shanghai China

ISBN: (纸本)9781450399029

Diabetic retinopathy (DR) has already been one of the leading causes of vision loss. A large number of researches about deep learning-based DR screening using color retinal photography images have been proposed in recent years. However, existing works mainly concentrated on non-proliferative diabetic retinopathy (NPDR) lesions. The exploration of proliferative diabetic retinopathy (PDR) lesion, which is more serious, is still insufficient. In this paper, we explored into one typical PDR lesion named neovascularization (NV). We collected and annotated retinal images from public and private datasets to construct and release a new dataset for NV detection. Three tasks including segmentation, detection, and grading were conducted on our dataset by using a set of state-of-the-art deep learning models. The results show that introducing NV detection can benefit DR grading. However, the segmentation and detection of NV are still challenging and have considerable room for improvement. © 2023 Copyright held by the owner/author(s).

关键词： Color

来源：评论

学校读者我要写书评

暂无评论

Detection of Steel Surface Defect Based on Faster R-CNN and FPN 2021

Detection of Steel Surface Defect Based on Faster R-CNN and ...

引用

7th International Conference on Computing and Artificial Intelligence, ICCAI 2021

作者： Yin, Tianping Yang, Jie Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University China

ISBN: (纸本)9781450389501

With the development of economy, steel has been widely used by human beings in various fields of social life, for example, military products, vehicles, aerospace, high-rise buildings and daily necessities. Therefore, it is very important to get high quality steel in industrial production. Detection of steel surface defect is a vital process in steel quality inspection. Traditional method uses artificial feature to detect steel surface defect, which costs lots of time to get features, and it is not robust in new environment. Therefore, this paper proposes an efficient and accurate method to solve the problem. In this paper, we adopt the faster R-CNN algorithm as baseline to train our own model. The feature pyramid network (FPN) is added to the original network of faster R-CNN, so that the network can combine high-level feature information and low-level feature information, furthermore, we replace the region-of-interest (RoI) pooling module with RoI align to reduce quantization error, which helps the mean average precision (mAP) increase about 0.8%. We use cycle GAN to do data enhancement, which helps our network easily to converge. Another problem is that it is quite common to get extreme aspect ratio in steel surface samples, which makes the detection more difficult, we introduce a method called multi-layer RoI align to solve this problem, which makes the mAP increase about 3.2%. According to our experiment results, the proposed method has a quite good performance on detection of steel surface defect. © 2021 ACM.

关键词： Aspect ratio

来源：评论

学校读者我要写书评

暂无评论

Revisiting the Benefits of Knowledge Distillation against Adversarial Examples

Revisiting the Benefits of Knowledge Distillation against Ad...

引用

Data Science and Information Technology (DSIT), International Conference on

作者： Jiawen Li Jie Yang Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China

Deep neural networks (DNNs) have always been a popular base model in many image classification tasks. However, some recent works suggest that there are some man made images will easily lead DNNs give wrong predictions. These images are generated by adding some invisible perturbations into clean images, which are the so-called adversarial examples. Since they are almost similar to original data but can fool DNNs, adversarial examples can bring the insecurity to many life-critical applications. Adversarial knowledge distillation is a positive defense method to attach improved robustness to the target DNNs against adversarial examples. In distillation, student models are trained by soft labels coming from robust pre-trained teacher networks. However, it is hard for researchers to ensure that supervised information comes from a better teacher model. In this paper, We focus on four potential factors of the teacher that may have an effect to the distillation performance of student model, and investigate whether and when they can boost the model robustness against adversarial attacks. Multiple quantitative experiment illustrate that these factors have impact on the distillation process and these results can be considered to choose a better teacher model. Further studies can focus on the combination of adversarial knowledge distillation and other defense schemes. Empirical results in this paper may also help select proper teacher in new complex distillation tasks.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Robust Deep Facial Attribute Prediction against Adversarial Attacks 2021

Robust Deep Facial Attribute Prediction against Adversarial ...

引用

7th International Conference on Computing and Artificial Intelligence, ICCAI 2021

作者： Fang, Kun Yang, Jie Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University China

ISBN: (纸本)9781450389501

Face recognition has always been a hot topic in research, and has also widely been applied in industry areas and daily life. Nowadays, face recognition models with excellent performance are mostly based on deep neural networks (DNN). However, recently researchers find that images added invisible perturbations could successfully fool neural networks, which is known as the so-called adversarial attack. The perturbed images, also known as adversarial examples, are almost the same as the original images, but neural network could give different and wrong predictions with high confidence on these adversarial examples. Such a phenomenon indicates the vulnerable robustness of neural network and thus casts a shadow on the security of DNN-based face recognition models. Therefore, in this paper, we focus on the facial attribute prediction task in face recognition, investigate the influence of adversarial attack on facial attribute prediction and give a solution on improving the robustness of facial attribute prediction models. Extensive experiment results illustrate that the solution could indeed produce much more robust results in facial attribute prediction against adversarial attacks. © 2021 ACM.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

GRC-Net: Grouping Guided R-band Chromosome recognition Network 24

GRC-Net: Grouping Guided R-band Chromosome Recognition Netwo...

引用

10th International Conference on Computing and Artificial Intelligence, ICCAI 2024

作者： Xia, Chao Liu, Wei Liu, Zhao Jiang Chen, Bing Yang, Jie Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China Shanghai Institute of Hematology State Key Laboratory of Medical Genomics National Research Center for Translational Medicine at Shanghai Ruijin Hospital Shanghai Jiao Tong University School of Medicine Shanghai China

ISBN: (纸本)9798400717055

Chromosome recognition is a critical and time-consuming process in karyotyping, especially for R-band chromosomes with poor visualization quality. In this paper, we propose an end-to-end grouping guided R-band chromosome recognition method GRC-Net. GRC-Net serves the chromosome recognition task as the main task and takes the chromosome length grouping task and centromere position grouping task as auxiliary tasks. Two auxiliary modules, Chromosome Length Grouping Module (CLGM) and Centromere Position Grouping Model (CPGM), are designed to extract the task-specific feature and refine the feature map of the main task. A large-scale R-band chromosome dataset with 1735 cases was collected. Experiment results on the 423 testing cases show that the proposed GRC-Net gets the highest accuracy of 96.87%, outperforming the baseline by 2.24%. With grouping task-guided feature extraction, GRC-Net reduces approximately 50% of the inter-group misclassification. The proposed GRC-Net can serve as a general framework for incorporating the domain knowledge into the process of feature learning, meanwhile, a powerful tool to assist clinical chromosome karyotyping. © 2024 ACM.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：