Deep Transfer Learning Based Multiway Feature Pyramid Network for Object Detection in Images

Object detection is being widely used in many fields, and therefore, the demand for more accurate and fast methods for object detection is also increasing. In this paper, we propose a method for object detection in digital images that is more accurate and faster. The proposed model is based on Single-Stage Multibox Detector (SSD) architecture. This method creates many anchor boxes of various aspect ratios based on the backbone network and multiscale feature network and calculates the classes and balances of the anchor boxes to detect objects at various scales. Instead of the VGG16-based deep transfer learning model in SSD, we have used a more efficient base network, i.e., EfficientNet. Detection of objects of different sizes is still an inspiring task. We have used Multiway Feature Pyramid Network (MFPN) to solve this problem. The input to the base network is given to MFPN, and then, the fused features are given to bounding box prediction and class prediction networks. Softer-NMS is applied instead of NMS in SSD to reduce the number of bounding boxes gently. The proposed method is validated on MSCOCO 2017, PASCAL VOC 2007, and PASCAL VOC 2012 datasets and compared to existing state-of-the-art techniques. Our method shows better detection quality in terms of mean Average Precision (mAP).

Accident vehicle types classification: a comparative study between different deep learning models

<span>Classifying and finding type of individual vehicles within an accident image are considered difficult problems. This research concentrates on accurately classifying and recognizing vehicle accidents in question. The aim to provide a comparative analysis of vehicle accidents. A number of network topologies are tested to arrive at convincing results and a variety of matrices are used in the evaluation process to identify the best networks. The best two networks are used with faster recurrent convolution neural network (Faster RCNN) and you only look once (YOLO) to determine which network will identifiably detect the location and type of the vehicle. In addition, two datasets are used in this research. In consequence, experiment results show that MobileNetV2 and ResNet50 have accomplished higher accuracy compared to the rest of the models, with 89.11% and 88.45% for the GAI dataset as well as 88.72% and 89.69% for KAI dataset, respectively. The findings reveal that the ResNet50 base network for YOLO achieved higher accuracy than MobileNetV2 for YOLO, ResNet50 for Faster RCNN with 83%, 81%, and 79% for GAI dataset and 79%, 78% and 74% for KAI dataset.</span>

A Plan-Do-Study-Act Approach to the Development, Implementation and Evaluation of a Patient Navigation Program to Reduce Breast Cancer Screening Disparities in Un- and Under-Insured, Racially and Ethnically Diverse Urban Women

Introduction: For the over 28 million Americans without health insurance, there is a great need to develop programs that help meet the health needs of the uninsured population.Materials and Methods: We applied the Plan-Do-Study-Act (PDSA) quality improvement framework to the development, implementation, and evaluation of a breast cancer screening navigation program for un- and under-insured women.Results: Six critical steps emerged: (1) obtain program funding; (2) navigator training; (3) establish a referral base network of community partners that serve the un- and under-insured women; (4) implement a process to address the barriers to accessing mammography; (5) develop a language- and culturally-tailored messaging and media campaign; and (6) develop measures and process evaluation to optimize and expand the program's reach.Discussion: A Plan-Do-Study-Act approach allowed identification of the key elements for successful development, implementation and optimization of a breast cancer screening navigation program aimed at reaching and screening un- and underinsured women.

Phase Ambiguity Resolution Method Using Processing of Code and Carrier Phase Pseudoranges in a Kalman-Type Filter

The main problem of implementing high-precision pseudoranges by carrier phase lies in their ambiguity associated with the ambiguity of the phase measurements of the navigation receiver. Thus, the development of new methods for phase ambiguity resolution becomes a very important element of high-precision positioning. The paper considers relative methods for estimating the coordinates of a stationary object that involve the use of both user and base (network in the case of a network of base receivers) receivers with precisely known coordinates located at a distance of several thousand kilometers from each other. We propose an algorithm for phase ambiguity resolution (integer type) based on the use of a Kalman-type filter (KTF), which receives ionosphere-free combinations of code and carrier phase pseudoranges. It is shown that traditional methods of ambiguity resolution require a significant observation period (about 2,000 seconds). We propose a method for evaluating the linear combination of phase ambiguities in the L1 and L2 bands obtained from instantaneous phase measurements. Its application along with the estimation of KTF parameters makes it possible to resolve phase ambiguities from as early as 50 seconds of observation. Set forth are the results of an experiment, in which code pseudorange measurements are used prior to the resolution of phase ambiguities and carrier phase pseudorange measurements are used after ambiguity resolution.

Unorganized point classification for robust NURBS surface reconstruction using a point-based neural network

In this paper, a method for classifying 3D unorganized points into interior and boundary points using a deep neural network is proposed. The classification of 3D unorganized points into boundary and interior points is an important problem in the nonuniform rational B-spline (NURBS) surface reconstruction process. A part of an existing neural network PointNet, which processes 3D point segmentation, is used as the base network model. An index value corresponding to each point is proposed for use as an additional property to improve the classification performance of the network. The classified points are then provided as inputs to the NURBS surface reconstruction process, and it has been demonstrated that the reconstruction is performed efficiently. Experiments using diverse examples indicate that the proposed method achieves better performance than other existing methods.

Phenomena Analysis of Ferroresonance and Self-Excitation in Subsea Power Systems

This article addresses the study regarding the emergence of ferroresonance and selfexcitation phenomena in Subsea Power Systems - SPS, composed essentially of synchronous generators installed on a platform (Topside), a three-phase umbilical cable and the electrical loads, the latter constituted by induction machines located on the seabed and connected to the umbilical through a power transformer and power electronic converters. Such phenomena are conceptually stated and characterized in the scope of SPS and the simulations are carried out in the PSCAD/EMTDC software, in its parallel processing environment, to verify indications of the existence of problems in the base network of the subsea distribution system.