With the rapid growth of fashion-focused social networks and online shopping, intelligent fashion recommendation is now in great needs. Recommending outfits, each which composed multiple interacted clothing accessories, relatively new to field. The problem becomes even more interesting challenging when considering users' personalized style. Another challenge a large-scale outfit system efficiency issue item/outfit search storage. In this paper, we propose learn binary code for efficient...

The classification of music genres has been a key focus in the field information retrieval (MIR), many researchers are also looking for number ways to solve this problem. Early approach Feature-based methods such as Support Vector Machines (SVMs) and K-nearest Neighbors (k-NN) mainly used handcrafted features Mel frequency cepstrum coefficients (MFCC). As convolutional neural networks (CNNs) increasingly capture complex time-frequency patterns audio data, performance accuracy genre...

The multimedia community has recently seen a tremendous surge of interest in the fashion recommendation problem. A lot efforts have been made to model compatibility between items. Some also studied users' personal preferences for outfits. There is, however, another difficulty task that hasn't dealt with carefully by previous work. Users are new system usually only several (less than 5) outfits available learning. With such limited number training examples, it is challenging user's reliably....

Passive non-line-of-sight (NLOS) imaging has drawn great attention in recent years. However, all existing methods are common limited to simple hidden scenes, low-quality reconstruction, and small-scale datasets. In this paper, we propose NLOS-OT, a novel passive NLOS framework based on manifold embedding optimal transport, reconstruct high-quality complicated scenes. NLOS-OT converts the high-dimensional reconstruction task low-dimensional mapping through alleviating ill-posedness imaging....

Infrared (IR) small target detection in sky scenes is crucial for aerospace, border security, and atmospheric monitoring. Most current works are typically designed generalized IR scenes, which may not be optimal the specific scenario of backgrounds, particularly detecting dim targets at long ranges. In these scenarios, presence heavy clouds usually causes significant false alarms due to factors such as strong edges, streaks, large undulations, isolated floating clouds. To address challenges,...

This paper demonstrates human synthesis based on the Radio Frequency (RF) signals, which leverages fact that RF signals can record movements with signal reflections off body. Different from existing sensing works only perceive humans roughly, this aims to generate fine-grained optical images by introducing a novel cross-modal RFGAN model. Specifically, we first build radio system equipped horizontal and vertical antenna arrays transceive signals. Since reflected are processed as obscure...

Abstract Despite the enormous potential of millimeter-wave (mmWave) imaging, high cost large-scale antenna arrays or stringent prerequisites synthetic aperture radar (SAR) principle impedes its widespread application. Here, we report a portable, affordable, and high-resolution 3D mmWave imaging system by overcoming destructive motion error handheld SAR imaging. This is achieved revealing two important phenomenons: spatial asymmetry errors in different directions, local similarity phase...

Cardiovascular diseases claim over 10 million lives annually, highlighting the critical need for long-term monitoring and early detection of cardiac abnormalities. Existing techniques like electrocardiograms (ECG) Holter are accurate but suffer from discomfort caused by body-attached electrodes. While wearable devices using photoplethysmography offer more convenience, they sacrifice accuracy susceptible to environmental interference. Here we present a radio frequency (RF)-based (60 64 GHz)...

Existing deep learning-based wireless sensing models usually require intensive computation. In this paper, we introduce a lightweight RF-based 3D human pose estimation model, i.e., Fast RFPose, to enable real-time estimation. Specifically, RFPose first estimates the locations in RF heatmap and crops location regions, then fine-grained poses based on cropped small heatmaps. experiments, build radio system multi-view camera acquire signals ground-truth poses, compare with state-of-the-art...

In order to analyze the performance of SC (Single-Carrier) 60 GHz communication system, a structure physical layer is designed based on IEEE 802.11ad standard. Supporting up 2.16 channel bandwidth, transmission rate close 7G bit/s. At same time, it adopt single carrier and OFDM (Orthogonal Frequency Division Multiplexing) technology. addition, using beam forming technology offset high path loss 60GHz frequency band, supporting fast session transfer between 2.4GHz, 5GHz band. Existing Wi-Fi...

Fashion outfit recommendation has attracted lots of attention recently. The problem becomes even more interesting and challenging when considering users’ personalized fashion preferences. Although existing works have successfully improved the accuracy, efficiency issue computation storage is still under-investigated often ignored. In this paper, we propose a discrete content-based tensor factorization model that maps items user to binary codes for efficient recommendation. We introduce...

In mammography, the efficacy of computer-aided detection methods depends, in part, on robust localisation micro-calcifications ($μ$C). Currently, most effective are based three steps: 1) individual $μ$C candidates, 2) clustering and 3) classification clusters. Where second step is motivated both to reduce number false positive detections from first evidence that malignancy depends a relatively large within certain area. this paper, we propose novel approach detection, consisting \emph{and}...

Existing RF-based human pose estimation methods usually require intensive computations and cannot meet the real-time processing portability requirements for mobile devices. To tackle limitation, in this article, we introduce a lightweight model, i.e., MobiRFPose, to construct portable camera. Different from traditional optical-based cameras, camera does not capture visual information, which means privacy-preserving characteristic. Specifically, only utilize horizontal antenna array...

In this paper, we introduce the outlier item detection task, which is related to compatibility prediction. Although, with ability of measuring compatibility, are able identify items that do not match overall style a given outfit, task has been well studied before. Most existing methods on prediction focus improving recommendation accuracy by utilizing underlying high order relationships among and have achieved promising results. Since these designed address above problem, performance can be...

This paper tackles a new challenge: how to generate human pose images from wireless signals? Although the optical camera can capture images, it is easily restricted by bad lighting. The signals do not rely on visible lights. However, low-resolution characteristics make previous works only rough skeleton of posture, missing lot detailed visual information, such as background, appearance, etc. Since information usually maintains unchanged for period and movements human, in this paper, we...

Millimeter-wave (mmWave) radar has gained increasing attention in environmental perception due to its robustness under low-light conditions. However, existing methods fail address the challenges of multipath interference and low angle resolution. In this paper, we introduce DiffRadar which leverages diffusion probabilistic model (DPM) for high-quality mmWave sensing. To adapt DPM signals that lack pix-level structural information, design a contour encoder capture intrinsic scene features...

Radar sensing has been a promising solution for contactless monitoring of Heart Rate Variability (HRV), an essential indicator the cardiovascular and autonomic nervous systems. However, existing works neglect heartbeat-driven body surface motions spreading across entire with spatial variations, which limits their accuracy in identifying fine-grid consecutive heartbeat timings overall HRV performance. In this paper, we propose to exploit reflections model inherent spatial-temporal...