Search

The proliferation of mobile devices is producing a new wave of mobile visual search applications that enable users to sense their surroundings with smart phones. As the particular challenges of mobile visual search, achieving high recognition bitrate becomes the consistent target of existed related works. In this paper, we explore to holistically exploit the deep learning-based hashing methods for more robust and instant mobile visual search. Firstly, we present a comprehensive survey of the existed deep learning based hashing methods, which showcases their remarkable power of automatic learning highly robust and compact binary code representation for visual search. Furthermore, in order to implement the deep learning hashing on computation and memory constrained mobile device, we investigate the deep learning optimization works to accelerate the computation and reduce the model size. Finally, we demonstrate a case study of deep learning hashing based mobile visual search system. The evaluations show that the proposed system can significantly improve 70% accuracy in MAP than traditional methods, and only needs less than one second computation time on the ordinary mobile phone. Finally, with the comprehensive study, we discuss the open issues and future research directions of deep learning hashing for mobile visual search. [ABSTRACT FROM AUTHOR]

The outbreak and rapid spread of coronavirus disease 2019 (COVID-19) has had a huge impact on the lives and safety of people around the world. Chest CT is considered an effective tool for the diagnosis and follow-up of COVID-19. For faster examination, automatic COVID-19 diagnostic techniques using deep learning on CT images have received increasing attention. However, the number and category of existing datasets for COVID-19 diagnosis that can be used for training are limited, and the number of initial COVID-19 samples is much smaller than the normal's, which leads to the problem of class imbalance. It makes the classification algorithms difficult to learn the discriminative boundaries since the data of some classes are rich while others are scarce. Therefore, training robust deep neural networks with imbalanced data is a fundamental challenging but important task in the diagnosis of COVID-19. In this paper, we create a challenging clinical dataset (named COVID19-Diag) with category diversity and propose a novel imbalanced data classification method using deep supervised learning with a self-adaptive auxiliary loss (DSN-SAAL) for COVID-19 diagnosis. The loss function considers both the effects of data overlap between CT slices and possible noisy labels in clinical datasets on a multi-scale, deep supervised network framework by integrating the effective number of samples and a weighting regularization item. The learning process jointly and automatically optimizes all parameters over the deep supervised network, making our model generally applicable to a wide range of datasets. Extensive experiments are conducted on COVID19-Diag and three public COVID-19 diagnosis datasets. The results show that our DSN-SAAL outperforms the state-of-the-art methods and is effective for the diagnosis of COVID-19 in varying degrees of data imbalance. [ABSTRACT FROM AUTHOR]

• We propose an improved dataset for research on image privacy. • We build a knowledge graph that represents the relevance between the objects and image privacy. • Based on the graph, we deal with the traditional image classification problem with the novel graph-based neural network. • The introduction of the knowledge graph not only makes the model more explainable but also makes better use of the information of objects provided by the images. • Our method relies on the intrinsic relevance in the dataset rather than extra knowledge, thus is applicable for other tasks. With the development of social media and smartphones, people share their daily lives via a large number of images, but the convince also raises a problem of privacy leakage. Therefore, effective methods are needed to infer the privacy risk of images and identify images that may disclose privacy. Several works have tried to solve this problem with deep learning models. However, we know little about how the models infer the privacy label of an image, thus it is not easy to understand why the image may disclose privacy. Inspired by recent research on graph neural networks, we introduce prior knowledge to the deep models to make the inference more explainable. We propose the Graph-based neural networks for Image Privacy (GIP) to infer the privacy risk of images. The GIP mainly focuses on objects in an image, and the knowledge graph is extracted from the objects in the dataset without reliance on extra knowledge. Experimental results show that the GIP achieves higher performance compared with the object-based methods and comparable performance even compared with the multi-modal fusion method. The results show that the introduction of the knowledge graph not only makes the deep model more explainable but also makes better use of the information of objects provided by the images. Combing the knowledge graph with deep learning is a promising way to help protect image privacy that is worth exploring. [ABSTRACT FROM AUTHOR]

Abstract Lung cancer is one of the leading causes of cancer-related death worldwide. Early diagnosis can effectively reduce the mortality, and computer-aided diagnosis (CAD) as an important way to assist doctors has developed rapidly. In particular, automated pulmonary nodule detection in computed tomography (CT) images is crucial to CAD. It is a challenging task to quickly locate the exact positions of lung nodules. In this paper, a novel automated pulmonary nodule detection framework with 2D convolutional neural network (CNN) is proposed to assist the CT reading process. Firstly, we adjust the structure of Faster R-CNN with two region proposal networks and a deconvolutional layer to detect nodule candidates, and then three models are trained for three kinds of slices for later result fusion. Secondly, a boosting architecture based on 2D CNN is designed for false positive reduction, which is a classifier to distinguish true nodules from the candidates. The misclassified samples are still kept for retraining a model which boosts the sensitivity for pulmonary nodule detection. Finally, the results of these networks are fused to vote out the final classification results. Extensive experiments are conducted on LUNA16, and the sensitivity of nodule candidate detection achieves 86.42%. For the false positive reduction, the sensitivity reaches 73.4% and 74.4% at 1/8 and 1/4 FPs/scan, respectively. It illustrates that the proposed method can obviously achieve accurate pulmonary nodule detection. [ABSTRACT FROM AUTHOR]