Search

In medical fields, data sharing for patients can improve the collaborative diagnosis and the complexity of traditional medical treatment process. Under the condition of data supervision, federated learning breaks the restrictions between medical institutions and realizes the sharing of medical data. However, there are still some issues. For example, lack of trust among medical institutions leads to the inability to establish safe and reliable cooperation mechanisms. For another example, malicious medical institutions destroy model aggregation by sharing false parameters. In this paper, we propose a new federated learning scheme based on blockchain architecture for medical data sharing. Moreover, we propose an intelligent contract to verify the identity of participants and detect malicious participants in federated learning. The experimental results show that the proposed data sharing scheme provides a credible participation mechanism for medical data sharing based on federal learning, and provides both higher efficiency and lower energy consumption as well.

Global average pooling (GAP) allows convolutional neural networks (CNNs) to localize discriminative information for recognition using only image-level labels. While GAP helps CNNs to attend to the most discriminative features of an object, e.g., head of a bird or one man's bag, it may suffer if that information is missing due to camera viewpoint changes and intraclass variations in some tasks. To circumvent this issue, we propose one new module to help CNNs to see more, namely, Spatial Rescaling (SpaRs) layer. It introduces spatial relations among the feature map activations back to the model, guiding the model to focus on a broad area in the feature map. With simple implementation, it can be inserted into CNNs of various architectures directly. SpaRs layer consistently improves the performance over the reidentification (re-ID) models. Besides, the new module based on different normalization methods also demonstrates the superiority of fine-grained and general image classification benchmarks. The visualization method shows the changes in activated regions when equipped with the SpaRs layer for better understanding. Our code is publicly available at https://github.com/HRanWang/Spatial-Re-Scaling.

Recently, automatic visual data understanding from drone platforms becomes highly demanding. To facilitate the study, the Vision Meets Drone Object Detection in Image Challenge is held the second time in conjunction with the 17-th International Conference on Computer Vision (ICCV 2019), focuses on image object detection on drones. Results of 33 object detection algorithms are presented. For each participating detector, a short description is provided in the appendix. Our goal is to advance the state-of-the-art detection algorithms and provide a comprehensive evaluation platform for them. The evaluation protocol of the VisDrone-DET2019 Challenge and the comparison results of all the submitted detectors on the released dataset are publicly available at the website: http: //www.aiskyeye.com/. The results demonstrate that there still remains a large room for improvement for object detection algorithms on drones.

Scale variation is one of the primary challenges in the object detection, existing in both inter-class and intra-class instances, especially on the drone platform. The latest methods focus on feature pyramid for detecting objects at different scales. In this work, we propose two techniques to refine multi-scale features for detecting various-scale instances in FPN-based Network. A Receptive Field Expansion Block (RFEB) is designed to increase the receptive field size for high-level semantic features, then the generated features are passed through a Spatial-Refinement Module (SRM) to repair the spatial details of multi-scale objects in images before summation by the lateral connection. To evaluate its effectiveness, we conduct experiments on VisDrone2019 benchmark dataset and achieve impressive improvement. Meanwhile, results on PASCAL VOC and MS COCO datasets show that our model is able to reach the competitive performance.

In the light detection and ranging (LiDAR) data classification, there are two problems affecting the LiDAR data classification performance. One is limited labeled samples in LiDAR dataset, another is confusing categories with extremely similar appearances. To address these challenges, this paper proposes a novel semi-supervised extended label algorithm (SSELA) for the data expansion, based on a novel pyramid feature co-training network (PFCTN). The proposed approach characterizes each level of feature pyramids without increasing the computational burden and time consumption, which has a benefit for learning complementary features of confusing categories. Experimental results demonstrate that the PFCTN alleviates the decreasing accuracy phenomenon at different scales of similar scenes, and achieves the promising results with limited labeled LiDAR dataset.

Most of the traditional line loss calculation methods are based on a period of time for line loss calculation, that is, the load current, voltage and other factors are calculated in an average manner over a period of time. In the selection of period, since the variation characteristics of the load itself are not taken into consideration, it lacks scientific basis. In this paper, the time-oriented process method is used to describe the change of load,and using hybrid clustering to divide time period,so the irregular load is divided into linear regular load. Finally, the line loss calculation is performed for each time period, which increases the theoretical accuracy of the line loss calculation and reduces the number of calculation.

Object detection is a hot topic with various applications in computer vision, e.g., image understanding, autonomous driving, and video surveillance. Much of the progresses have been driven by the availability of object detection benchmark datasets, including PASCAL VOC, ImageNet, and MS COCO. However, object detection on the drone platform is still a challenging task, due to various factors such as view point change, occlusion, and scales. To narrow the gap between current object detection performance and the real-world requirements, we organized the Vision Meets Drone (VisDrone2018) Object Detection in Image challenge in conjunction with the 15th European Conference on Computer Vision (ECCV 2018). Specifically, we release a large-scale drone-based dataset, including 8, 599 images (6, 471 for training, 548 for validation, and 1, 580 for testing) with rich annotations, including object bounding boxes, object categories, occlusion, truncation ratios, etc. Featuring a diverse real-world scenarios, the dataset was collected using various drone models, in different scenarios (across 14 different cities spanned over thousands of kilometres), and under various weather and lighting conditions. We mainly focus on ten object categories in object detection, i.e., pedestrian, person, car, van, bus, truck, motor, bicycle, awning-tricycle, and tricycle. Some rarely occurring special vehicles (e.g., machineshop truck, forklift truck, and tanker) are ignored in evaluation. The dataset is extremely challenging due to various factors, including large scale and pose variations, occlusion, and clutter background. We present the evaluation protocol of the VisDrone-DET2018 challenge and the comparison results of 38 detectors on the released dataset, which are publicly available on the challenge website: http://www.aiskyeye.com/. We expect the challenge to largely boost the research and development in object detection in images on drone platforms.

Knowledge of phase boundaries in multi-component system plays a critical role in exploring the alloy design space effectively and safely, which defines the roadmap to stabilize desirable phases and sidestep detrimental phases. In this work, a novel combinatorial approach was developed to determine the phase boundaries in multi-component nickel-based superalloys. In addition, a three-phase region was pinpointed and expected to enhance high-temperature properties. The results were further analyzed to evaluate the feasibility of CALPHAD and semi-empirical predictions, which provided clues for the improvement of the thermodynamic databases.

Yield strength (YS) is a key factor during design and application of Ni-based superalloys with complex compositions, hence it is of great significance to evaluate the YS prior to manufacturing. In this work, alloy diffusion-multiple technology was employed as a high-throughput way to yield the hardness dataset. Based on the composition and other descriptors, Pearson correlation coefficients, stability selection and feature importance were used to select the efficient feature variables. Thereafter, six different machine learning models were applied to predict the YS. Finally, the individual and interaction effect of Co and Mo could be effectively detected by the Gaussian process regression (GPR) model. The optimum composition of Ni-based superalloys with the largest YS at room temperature was determined using the trained GPR model and genetic algorithm. This method can be extended to predict the YS in other multicomponent alloys, such as Ti alloys, Co-based alloys, and high entropy alloys.