Lundgrenjansen1440

Thirdly, as for the local-local alignment, we match visual human parts with noun phrases in the Bi-directional Fine-grained Matching (BFM) module. The whole network combining multiple granularities can be end-to-end trained without complex preprocessing. To address the difficulties in training the combination of multiple granularities, an effective step training strategy is proposed to train these granularities step-by-step. Extensive experiments and analysis have shown that our method obtains the state-of-the-art performance on the CUHK-PEDES dataset and outperforms the previous methods by a significant margin.Robust spatiotemporal representations of natural videos have several applications including quality assessment, action recognition, object tracking etc. In this paper, we propose a video representation that is based on a parameterized statistical model for the spatiotemporal statistics of mean subtracted and contrast normalized (MSCN) coefficients of natural videos. Specifically, we propose an asymmetric generalized Gaussian distribution (AGGD) to model the statistics of MSCN coefficients of natural videos and their spatiotemporal Gabor bandpass filtered outputs. We then demonstrate that the AGGD model parameters serve as good representative features for distortion discrimination. Based on this observation, we propose a supervised learning approach using support vector regression (SVR) to address the no-reference video quality assessment (NRVQA) problem. The performance of the proposed algorithm is evaluated on publicly available video quality assessment (VQA) datasets with both traditional and in-capture/authentic distortions. We show that the proposed algorithm delivers competitive performance on traditional (synthetic) distortions and acceptable performance on authentic distortions. The code for our algorithm will be released at https//www.iith.ac.in/~lfovia/downloads.html.Spatial resolution in conventional sonography is achieved through focusing and steering of an ultrasound beam. However, due to acoustic diffraction, the ability to focus an ultrasound beam is limited which leads to low spatial and contrast resolutions. We aim to propose a new method wherein the array elements are simultaneously excited with signals coded with random sequences, which yields an unfocused ultrasound wavefront of random interference. When such a wavefront propagates through the medium, its energy reflects back from the tissue, causing individual scatterers to have unique impulse responses. In such a case, we can reconstruct high-resolution ultrasound images using a priori measurements of spatial impulse responses and the l1 norm minimization algorithm. Selleck Darolutamide In a simulation study, we achieved a spatial resolution of 0.25 mm, which constitutes a four-fold improvement over conventional methods that use delay-and-sum beamforming. In the experimental study, we demonstrate the accuracy of the proposed interference-based method using a tissue-mimicking phantom with 0.1-mm- and 0.08- mm-diameter nylon wires.Current methods for in-vivo microvascular imaging (7.5MHz) are able to visualize smaller vasculature, however are still limited in the segmentation of lower velocity blood flow from moving tissue. Contrast enhanced ultrasound (CEUS) has been successful in visualizing changes in microvascular flow at conventional diagnostic ultrasound imaging frequencies ( less then 7.5MHz). However, conventional CEUS approaches at elevated frequencies have met with limited success, due inpart to the diminishing microbubble response with frequency. We apply a plane-wave acquisition combined with non-linear Doppler processing of ultrasound contrast agents at 15MHz to improve resolution of microvascular blood flow, while compensating for reduced microbubble response. This planewave Doppler approach of imaging ultrasound contrast agents also enables simultaneous detection and separation of blood flow in the microcirculation and higher velocity flow in the larger vasculature. We apply singular value decomposition filtering on the non-linear Doppler signal to orthogonally separate the more stationary lower velocity flow in the microcirculation and higher velocity flow in the larger vasculature. This orthogonal separation was also utilized to improve time intensity curve analysis of the microcirculation, by removing higher velocity flow corrupting bolus kinetics. We demonstrate the utility of this imaging approach in a rat spinal cord injury model, requiring sub-millimeter resolution.Based on a Ba(1-x)SrxTiO3 ferroelectric thin film, a discrete tunable surface mounted device (SMD) capacitor has been developed for microwave frequency applications purpose. The proposed SMD topology has the particular advantage of inherent decoupling between the RF signal and the DC biasing voltage, necessary to tune the ferroelectric permittivity. The design and technological development of the SMD component is presented and synthesis of the ferroelectric thin film is summarized. Material characterization shows convenient tunability, while low dielectric losses at 10 MHz. Integration of the SMD tunable capacitor into a Planar Inverted-F Antenna has been done in order to evaluate the agility and tunability performance of the antenna.Automatic Non-rigid Histological Image Registration (ANHIR) challenge was organized to compare the performance of image registration algorithms on several kinds of microscopy histology images in a fair and independent manner. We have assembled 8 datasets, containing 355 images with 18 different stains, resulting in 481 image pairs to be registered. Registration accuracy was evaluated using manually placed landmarks. In total, 256 teams registered for the challenge, 10 submitted the results, and 6 participated in the workshop. Here, we present the results of 7 well-performing methods from the challenge together with 6 well-known existing methods. The best methods used coarse but robust initial alignment, followed by non-rigid registration, used multiresolution, and were carefully tuned for the data at hand. They outperformed off-the-shelf methods, mostly by being more robust. The best methods could successfully register over 98 % of all landmarks and their mean landmark registration accuracy (TRE) was 0.44 % of the image diagonal.