基于编码器-解码器的半监督图像语义分割

基于深度卷积神经网络的图像语义分割方法需要大量像素级标注的训练数据,但标注的过程费时又费力.本文基于生成对抗网络提出一种编码-解码结构的半监督图像语义分割方法,其中编码器-解码器模块作为生成器,整个网络通过耦合标准多分类交叉熵损失和对抗损失进行训练.为充分利用浅层网络包含的丰富的语义信息,本文将编码器中不同尺度的特征输入到分类器,并将得到的不同粒度的分类结果融合,进而优化目标边界.此外,鉴别器通过发现无标签数据分割结果中的可信区域,以此提供额外的监督信号,来实现半监督学习.在PASCAL VOC 2012和Cityscapes上的实验表明,本文提出的方法优于现有的半监督图像语义分割方法.     

基于多任务模型的深度预测算法研究

图像的深度值预测是计算机视觉和机器人领域中的一个热门的研究课题.深度图的构建是三维重建的重要前提,传统方法主要依靠确定固定点深度进行人工标注或是根据相机的位置变化来进行双目定位预测深度,但这类方法一方面费时费力,另一方面也受到相机位置、定位方式、分布概率性等因素的限制,准确率很难得到保证,从而导致预测的深度图难以完成后...     

Learning multi-task local metrics for image annotation

The goal of image annotation is to automatically assign a set of textual labels to an image to describe the visual contents thereof. Recently, with the rapid increase in the number of web images, nearest neighbor (NN) based methods have become more attractive and have shown exciting results for image annotation. One of the key challenges of these methods is to define an appropriate similarity measure between images for neighbor selection. Several distance metric learning (DML) algorithms derived from traditional image classification problems have been applied to annotation tasks. However, a fundamental limitation of applying DML to image annotation is that it learns a single global distance metric over the entire image collection and measures the distance between image pairs in the image-level. For multi-label annotation problems, it may be more reasonable to measure similarity of image pairs in the label-level. In this paper, we develop a novel label prediction scheme utilizing multiple label-specific local metrics for label-level similarity measure, and propose two different local metric learning methods in a multi-task learning (MTL) framework. Extensive experimental results on two challenging annotation datasets demonstrate that 1) utilizing multiple local distance metrics to learn label-level distances is superior to using a single global metric in label prediction, and 2) the proposed methods using the MTL framework to learn multiple local metrics simultaneously can model the commonalities of labels, thereby facilitating label prediction results to achieve state-of-the-art annotation performance.     

面向多模态MRI脑胶质瘤区域三维分割与生存期预测的级联U-Net网络

目的 针对常见方法对脑胶质瘤的肿瘤分割和生存预测需要单独建模的问题,提出一种带有变分自编码器(variational auto-encoder, VAE)分支的两阶段级联U-Net算法,旨在分割肿瘤的同时提取鲁棒的特征预测患者生存期,有助于患者的精准治疗。方法 提出的两阶段级联U-Net网络,第1阶段实现初步粗分割,第2阶段实现精细化分割。此外,在第2阶段添加变分自编码器分支以提取更加鲁棒的特征并提高模型泛化性。其中,变分自编码器分支获取的特征被送入随机森林算法以进行生存期预测。另外,在两个阶段的解码器部分都添加了SE(squeeze-and-excitation)-残差模块以及注意力门模块,提高了分割精度。结果 在Brain Tumor Segmentation (BraTS)竞赛官网分别评估了本文方法在BraTS2020验证集上分割以及总体生存期预测两大任务的结果,本文算法在该验证集的全肿瘤区域、肿瘤核心区域以及增强型肿瘤区域分别取得了90.66%、85.09%和79.02%的Dice相似系数。相较3DU-Net在3个肿瘤子区域的Dice相似系数分别提高了4.3%、1.37%和5....     

基于双模型交互学习的半监督医学图像分割

在医学图像中, 器官或病变区域的精准分割对疾病诊断等临床应用有着至关重要的作用, 然而分割模型的训练依赖于大量标注数据. 为减少对标注数据的需求, 本文主要研究针对医学图像分割的半监督学习任务. 现有半监督学习方法广泛采用平均教师模型, 其缺点在于, 基于指数移动平均(Exponential moving average, EMA)的参数更新方式使得老师模型累积学生模型的错误知识. 为避免上述问题, 提出一种双模型交互学习方法, 引入像素稳定性判断机制, 利用一个模型中预测结果更稳定的像素监督另一个模型的学习, 从而缓解了单个模型的错误经验的累积和传播. 提出的方法在心脏结构分割、肝脏肿瘤分割和脑肿瘤分割三个数据集中取得优于前沿半监督方法的结果. 在仅采用30%的标注比例时, 该方法在三个数据集上的戴斯相似指标(Dice similarity coefficient, DSC)分别达到89.13%, 94.15%, 87.02%.     

拉普拉斯阶梯网络

阶梯网络不仅是一种基于深度学习的特征提取器,而且能够应用于半监督学习中.深度学习在实现了复杂函数逼近的同时,也缓解了多层神经网络易陷入局部最小化的问题.传统的自编码、玻尔兹曼机等方法易忽略高维数据的低维流形结构信息,使用这些方法往往会获得无意义的特征表示,这些特征不能有效地嵌入到后续的预测或识别任务中.从流形学习的角度出发,提出一种基于阶梯网络的深度表示学习方法,即拉普拉斯阶梯网络LLN(Laplacian ladder network).拉普拉斯阶梯网络在训练的过程中不仅对每一编码层嵌入噪声并进行重构,而且在各重构层引入图拉普拉斯约束,将流形结构嵌入到多层特征学习中,以提高特征提取的鲁棒性和判别性.在有限的有标签数据情况下,拉普拉斯阶梯网络将监督学习损失和非监督损失融合到了统一的框架进行半监督学习.在标准手写数据数据集MNIST和物体识别数据集CIFAR-10上进行了实验,结果表明,相对于阶梯网络和其他半监督方法,拉普拉斯阶梯网络都得到了更好的分类效果,是一种有效的半监督学习算法.     

Interactive image segmentation with a regression based ensemble learning paradigm

To achieve fine segmentation of complex natural images, people often resort to an interactive segmentation paradigm, since fully automatic methods often fail to obtain a result consistent with the ground truth. However, when the foreground and background share some similar areas in color, the fine segmentation result of conventional interactive methods usually relies on the increase of manual labels. This paper presents a novel interactive image segmentation method via a regression-based ensemble model with semi-supervised learning. The task is formulated as a non-linear problem integrating two complementary spline regressors and strengthening the robustness of each regressor via semi-supervised learning. First, two spline regressors with a complementary nature are constructed based on multivariate adaptive regression splines (MARS) and smooth thin plate spline regression (TPSR). Then, a regressor boosting method based on a clustering hypothesis and semi-supervised learning is proposed to assist the training of MARS and TPSR by using the region segmentation information contained in unlabeled pixels. Next, a support vector regression (SVR) based decision fusion model is adopted to integrate the results of MARS and TPSR. Finally, the GraphCut is introduced and combined with the SVR ensemble results to achieve image segmentation. Extensive experimental results on benchmark datasets of BSDS500 and Pascal VOC have demonstrated the effectiveness of our method, and the comparison with experiment results has validated that the proposed method is comparable with the state-of-the-art methods for interactive natural image segmentation.     

面向手术器械语义分割的半监督时空Transformer网络

基于内窥镜的微创手术机器人在临床上的应用日益广泛,为医生提供内窥镜视频中精准的手术器械分割信息,对提高医生操作的准确度、改善患者预后有重要意义.现阶段,深度学习框架训练手术器械分割模型需要大量精准标注的术中视频数据,然而视频数据标注成本较高,在一定程度上限制了深度学习在该任务上的应用.目前的半监督方法通过预测与插帧,可...     

Semi-supervised learning combining co-training with active learning

Co-training is a good paradigm of semi-supervised, which requires the data set to be described by two views of features. There are a notable characteristic shared by many co-training algorithm: the selected unlabeled instances should be predicted with high confidence, since a high confidence score usually implies that the corresponding prediction is correct. Unfortunately, it is not always able to improve the classification performance with these high confidence unlabeled instances. In this paper, a new semi-supervised learning algorithm was proposed combining the benefits of both co-training and active learning. The algorithm applies co-training to select the most reliable instances according to the two criterions of high confidence and nearest neighbor for boosting the classifier, also exploit the most informative instances with human annotation for improve the classification performance. Experiments on several UCI data sets and natural language processing task, which demonstrate our method achieves more significant improvement for sacrificing the same amount of human effort.     

体育视频中的运动员检测与分割

运动员的动作行为分析是体育视频高级语义分析的直接途径,检测和分割视频中的运动员是分析运动员的动作行为的基础.利用体育视频的领域规则和中层特征块的性质,提出半监督的方法挖掘运动员的中层特征块,针对不同类型的镜头分别训练基于中层特征块的运动员检测分类器,实现运动员检测.利用运动员检测结果标记超像素,结合Grab Cut分割算法实现运动员分割.实验结果表明,基于中层特征块的运动员区域检测算法能够快速挖掘训练所需样本,从而训练得到检测分类器,检测结果具有较高的准确度,获得的运动员区域能够有效用于运动员分割,简化了分割计算过程.     

基于半监督协同训练的百科知识库实体对齐

针对传统实体对齐方法中的有监督学习算法依赖大量标注数据,以及特征表示不适用于百科知识库等问题,提出一种基于半监督协同训练的实体对齐方法。将实体对齐建模为一个带约束的二分类问题,充分利用实体名、属性、描述文本及其中的时间、数值等关键信息,组合生成多维特征;将特征划分为2个相对独立的视图,通过2个视图上分类器的协同训练,迭代地从未标注数据中学习同义实体的分布情况。在2个中文百科上的实验结果表明,使用半监督协同训练方法进行实体对齐的F1值达到84.3%,较其他方法效果最优,证明了其有效性和在百科知识库上的实用价值。     

URO-GAN: An untrustworthy region optimization approach for adipose tissue segmentation based on adversarial learning

Automatic segmentation of adipose tissue from CT images is an essential module of medical assistant diagnosis. A large scale of abdominal cross-section CT images can be used to segment subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) with deep learning method. However, the CT images still need to be professionally and accurately annotated to improve the segmentation quality. The paper proposes a semi-supervised segmentation network based on adversarial learning. The model is called URO-GAN and consists of two paths used to segment SAT and VAT, respectively. An SAT-to-VAT transmission mechanism is set up between these two paths, where several inverse-SAT excitation blocks are set to help the SAT segmentation network guide the VAT segmentation network. An untrustworthy region optimization mechanism is proposed to improve the segmentation quality and keep the adversarial learning stable. With the confidence map output from the discriminator network, an optimizer network is used to fix the error in the masks predicted by the segmentation network. The URO-GAN achieves good results by training with 84 annotated images and 3969 unannotated images. Experimental results demonstrate the effectiveness of our approach on the segmentation of adipose tissue in medical images.

     

轮廓指导的层级混合多任务全卷积网络

传统的深度多任务网络通常在不同任务之间共享网络的大部分层(即特征表示层).由于这样做会忽视不同任务各自的特殊性,所以往往会制约其适应数据的能力.提出了一种层级混合的多任务全卷积网络HFFCN,以解决CT图像中的前列腺分割问题.特别地,使用一个多任务框架来解决这个问题.这个框架包括一个分割前列腺的主任务和一个回归前列腺边界的辅助任务.这里,第2个任务主要是用来精确地描述在CT图像中模糊的前列腺边界.因此,HFFCN架构是一个双分支的结构,包含一个编码主干和两个解码分支.不同于传统的多任务网络,提出了一个信息共享模块,用以在两个解码分支之间共享信息.这使得HFFCN可以学习任务的通用层级信息,同时保留一些不同任务各自的特征表示.在一个包含有313个病人的313张计划阶段图片的CT图像数据集上做了详细的实验.实验结果证明了HFFCN网络可以超越现有其他先进的分割方法或者传统的多任务学习模型.     

Weakly- and Semi-Supervised Fast Region-Based CNN for Object Detection

Learning an effective object detector with little supervision is an essential but challenging problem in computer vision applications. In this paper, we consider the problem of learning a deep convolutional neural network (CNN) based object detector using weakly-supervised and semi-supervised information in the framework of fast region-based CNN (Fast R-CNN). The target is to obtain an object detector as accurate as the fully-supervised Fast R-CNN, but it requires less image annotation effort. To solve this problem, we use weakly-supervised training images (i.e., only the image-level annotation is given) and a few proportions of fully-supervised training images (i.e., the bounding box level annotation is given), that is a weakly- and semi-supervised (WASS) object detection setting. The proposed solution is termed as WASS R-CNN, in which there are two main components. At first, a weakly-supervised R-CNN is firstly trained; after that semi-supervised data are used for finetuning the weakly-supervised detector. We perform object detection experiments on the PASCAL VOC 2007 dataset. The proposed WASS R-CNN achieves more than 85% of a fully-supervised Fast R-CNN’s performance (measured using mean average precision) with only 10% of fully-supervised annotations together with weak supervision for all training images. The results show that the proposed learning framework can significantly reduce the labeling efforts for obtaining reliable object detectors.

     

基于采样的半监督支持向量机软件缺陷预测方法

软件缺陷预测有助于提高软件开发质量,保证测试资源有效分配。针对软件缺陷预测研究中类标签数据难以获取和类不平衡分布问题,提出基于采样的半监督支持向量机预测模型。该模型采用无监督的采样技术,确保带标签样本数据中缺陷样本数量不会过低,使用半监督支持向量机方法,在少量带标签样本数据基础上利用无标签数据信息构建预测模型;使用公开的NASA软件缺陷预测数据集进行仿真实验。实验结果表明提出的方法与现有半监督方法相比,在综合评价指标[F]值和召回率上均优于现有方法;与有监督方法相比,能在学习样本较少的情况下取得相当的预测性能。     

多任务实时声音事件检测卷积模型与复合数据扩增

现有的声音事件检测研究多为对离线音频进行分析,且模型参数量较多、计算效率低,不适用于实时检测。提出一种面向多任务实时声音事件检测的轻量化卷积神经网络模型,它将唤醒与检测任务整合成多任务学习框架,此外模型的卷积结构联合了稠密连接、Ghost模组与SE注意力机制;另外还提出了一种复合数据扩增方法,将音频变换、随机裁剪与频谱掩蔽相结合。实验结果显示,该模型在ESC-10和Urbansound8K数据集上的平均预测准确率高于当前新型的基线模型2%以上,同时模型的参数和内存更少。研究表明,多任务学习的方式节省了计算量,又因为卷积结构复用了中间层特征,模型可以快速地反馈检测结果。另外,复合数据方法相比传统方法使模型获得了更好的性能和鲁棒性。     

A feature-free and parameter-light multi-task clustering framework

The two last decades have witnessed extensive research on multi-task learning algorithms in diverse domains such as bioinformatics, text mining, natural language processing as well as image and video content analysis. However, all existing multi-task learning methods require either domain-specific knowledge to extract features or a careful setting of many input parameters. There are many disadvantages associated with prior knowledge requirements for feature extraction or parameter-laden approaches. One of the most obvious problems is that we may find a wrong or non-existent pattern because of poorly extracted features or incorrectly set parameters. In this work, we propose a feature-free and parameter-light multi-task clustering framework to overcome these disadvantages. Our proposal is motivated by the recent successes of Kolmogorov-based methods on various applications. However, such methods are only defined for single-task problems because they lack a mechanism to share knowledge between different tasks. To address this problem, we create a novel dictionary-based compression dissimilarity measure that allows us to share knowledge across different tasks effectively. Experimental results with extensive comparisons demonstrate the generality and the effectiveness of our proposal.     

Semi-supervised kernel density estimation for video annotation

Insufficiency of labeled training data is a major obstacle for automatic video annotation. Semi-supervised learning is an effective approach to this problem by leveraging a large amount of unlabeled data. However, existing semi-supervised learning algorithms have not demonstrated promising results in large-scale video annotation due to several difficulties, such as large variation of video content and intractable computational cost. In this paper, we propose a novel semi-supervised learning algorithm named semi-supervised kernel density estimation (SSKDE) which is developed based on kernel density estimation (KDE) approach. While only labeled data are utilized in classical KDE, in SSKDE both labeled and unlabeled data are leveraged to estimate class conditional probability densities based on an extended form of KDE. It is a non-parametric method, and it thus naturally avoids the model assumption problem that exists in many parametric semi-supervised methods. Meanwhile, it can be implemented with an efficient iterative solution process. So, this method is appropriate for video annotation. Furthermore, motivated by existing adaptive KDE approach, we propose an improved algorithm named semi-supervised adaptive kernel density estimation (SSAKDE). It employs local adaptive kernels rather than a fixed kernel, such that broader kernels can be applied in the regions with low density. In this way, more accurate density estimates can be obtained. Extensive experiments have demonstrated the effectiveness of the proposed methods.     

融合跨阶段深度学习的脑肿瘤MRI图像分割

目的 磁共振成像(magnetic resonance imaging, MRI)作为一种非侵入性的软组织对比成像方式,可以提供有关脑肿瘤的形状、大小和位置等有价值的信息,是用于脑肿瘤患者检查的主要方法,在脑肿瘤分割任务中发挥着重要作用。由于脑肿瘤本身复杂多变的形态、模糊的边界、低对比度以及样本梯度复杂等问题,导致高精度脑肿瘤MRI图像分割非常具有挑战性,目前主要依靠专业医师手动分割,费时且可重复性差。对此,本文提出一种基于U-Net的改进模型,即CSPU-Net(cross stage partial U-Net)脑肿瘤分割网络,以实现高精度的脑肿瘤MRI图像分割。方法 CSPU-Net在U-Net结构的上下采样中分别加入两种跨阶段局部网络结构(cross stage partial module, CSP)提取图像特征,结合GDL(general Dice loss)和WCE(weighted cross entropy)两种损失函数解决训练样本类别不平衡问题。结果 在BraTS (brain tumor segmentation) 2018和BraTS 2019两个数据集上进行实...     

Automatic segmentation of optic disc in retinal fundus images using semi-supervised deep learning

Diseases of the eye require manual segmentation and examination of the optic disc by ophthalmologists. Though, image segmentation using deep learning techniques is achieving remarkable results, it leverages on large-scale labeled datasets. But, in the field of medical imaging, it is challenging to acquire large labeled datasets. Hence, this article proposes a novel deep learning model to automatically segment the optic disc in retinal fundus images by using the concepts of semi-supervised learning and transfer learning. Initially, a convolutional autoencoder (CAE) is trained to automatically learn features from a large number of unlabeled fundus images available from the Kaggle’s diabetic retinopathy (DR) dataset. The autoencoder (AE) learns the features from the unlabeled images by reconstructing the input images and becomes a pre-trained network (model). After this, the pre-trained autoencoder network is converted into a segmentation network. Later, using transfer learning, the segmentation network is trained with retinal fundus images along with their corresponding optic disc ground truth images from the DRISHTI GS1 and RIM-ONE datasets. The trained segmentation network is then tested on retinal fundus images from the test set of DRISHTI GS1 and RIM-ONE datasets. The experimental results show that the proposed method performs on par with the state-of-the-art methods achieving a 0.967 and 0.902 dice score coefficient on the test set of the DRISHTI GS1 and RIM-ONE datasets respectively. The proposed method also shows that transfer learning and semi-supervised learning overcomes the barrier imposed by the large labeled dataset. The proposed segmentation model can be used in automatic retinal image processing systems for diagnosing diseases of the eye.