Crowd counting is a new frontier in computer vision with far-reaching applications particularly social safety management. A majority of existing works adopt methodology that first estimates person-density map and then calculates integral over this to obtain the final count. As noticed by several prior investigations, learned density can significantly deviate from true person even though reported count precise. This implies unreliable for localizing crowd. To address issue, work proposes...

High-density semiconducting aligned carbon nanotube (A-CNT) arrays have been demonstrated with wafer-scale preparation of materials and shown high performance in P-type field-effect transistors (FETs) great potential for applications future digital integrated circuits (ICs). However, high-performance N-type FETs (N-FETs) not yet implemented A-CNTs, making development complementary metal-oxide-semiconductor (CMOS) technology, a necessary component modern ICs, impossible. In this work, we...

The motivation for this review paper came from the developing countries where economy is mostly dependent on agriculture and climate conditions. Based current conditions historical records, profitability in production farming depends making a right timely operational decision. Precision systematic program designed to maximize productivity of by carefully tailoring soil crop management meet specific requirements each field while preserving environmental quality. This highlights development an...

Video visual relation detection (VidVRD) aims to describe all interacting objects in a video. Different from relationships static images, videos contain an addition temporal channel. A majority of existing works divide video into short segments, predict each segment, and merge them. Such methods cannot capture relations involving long motions. Predicting the same relationship across neighboring segments is also inefficient. To address these issues, this work proposes novel sliding-window...

Nowadays, 2D barcodes have been widely used as an interface to connect potential customers and advertisement contents. However, the appearance of a conventional barcode pattern is often too obtrusive for integrating into aesthetically designed advertisement. Besides, no human readable information provided before successfully decoded. This paper proposes new picture-embedding barcode, called PiCode, which mitigates these two limitations by equipping scannable with picturesque appearance....

The biological behavior of carcinoma the esophagogastric junction (CEGJ) is different from that gastric or esophageal cancer. Differentiating squamous cell (SCCEG) adenocarcinoma (AEG) can indicate Siewert stage and whether surgical route for patients with CEGJ transthoracic transabdominal, as well aid in determining extent lymph node dissection. With development neoadjuvant therapy, preoperative determination pathological type help selection radiotherapy chemotherapy regimens.To establish...

An aligned semiconducting carbon nanotube (A-CNT) array has been considered an excellent channel material to construct high-performance field-effect transistors (FETs) and integrated circuits (ICs). The purification assembly processes prepare a A-CNT require conjugated polymers, introducing stubborn residual polymers stress at the interface between A-CNTs substrate, which inevitably affects fabrication performance of FETs. In this work, we develop process refresh Si/SiO2 substrate surface...

Wafer-scale aligned carbon nanotubes (A-CNTs) are promising candidate semiconductors for building high-performance complementary metal-oxide-semiconductor (CMOS) transistors future integrated circuits (ICs). A-CNT-based p-type field-effect (P-FETs) have demonstrated excellent performance and scalability down to sub-10 nm nodes. However, the development of A-CNT n-type FETs (N-FETs) lags far behind, in regard their electronic device scaling. In this work, we fabricated top-gated N-FETs based...

One popular hypothesis of neural network generalization is that the flat local minima loss surface in parameter space leads to good generalization. However, we demonstrate has no obvious relationship with generalization, especially under adversarial settings. Through visualizing decision surfaces both and input space, instead show geometry property correlates well robustness. We then propose an robustness indicator, which can evaluate a network's intrinsic without testing its accuracy...

This research presents the new techniques and practical experiences of using unmanned aerial vehicles (UAVs) precision agriculture mapping. UAV-based remote sensing systems should be cost-effective, fast-producing, have high geometric accuracy, simple to operate by local staff. work aims to: (1) precisely use high-resolution UAV thermal multi-spectral sensors machine learning approaches reliably assess crop water status on a field scale; (2) capture on-field images for quantitative study...

Metallic wafer bonding is becoming a key enabling technology in microelectromechanical systems packaging and heterogeneous integration. We realize the of silicon (Si) wafers coated with aluminum (Al) film thin tin (Sn) as intermediate layer. The pressure 0.25 MPa. achieved at temperatures low 280°C after short time 3 min. average strength 9.9 minimum variation layer thickness about 0.2 μm within large area. A fracture surface study cross-section analysis are conducted bond mechanism...

Abstract Objectives To observe and compare the difference in retinal thickness using optical coherence tomography (OCT) between patients with high myopia (HM) during third trimester of pregnancy age-matched HM non-pregnant women. Methods A case-control study. total 39 eyes from women (study group) 50 (control were included. All subjects underwent SD-OCT examination. The built-in software was used to measure macular region. data two groups compared independent-samples t test. Results Among 89...

Visual relation detection (VRD) aims to describe all interacting objects in an image using subject-predicate-object triplets. Critically, valid relations combinatorially grow O(C2 R) for C object categories and R relationships. The frequencies of triplets exhibit a long-tailed distribution, which inevitably leads bias towards popular visual the learned VRD model. To address this problem, we propose localize-assemble-predicate network (LAP-Net), decomposes into three sub-tasks: localizing...

The purpose of this work is to develop a 3-channel endorectal coil (ERC-3C) structure obtain higher signal-to-noise (SNR) and better parallel imaging performance for prostate magnetic resonance (MRI) at 3T.The was validated by in vivo studies the SNR, g-factor, diffusion-weighted (DWI) were compared. A 2-channel (ERC-2C) with two orthogonal loops 12-channel external surface employed comparison.Compared ERC-2C quadrature configuration array, proposed ERC-3C improved SNR 23.9% 428.9%,...

3-D human pose estimation is a crucial step for understanding actions. However, reliably capturing precise position of joints non-trivial and tedious. Current models often suffer from the scarcity high-quality annotated training data. In this work, we explore novel way obtaining gigantic data without manual annotations. catedioptric videos (\emph{e.g.}, people dance before mirror), camera records both original mirrored poses, which provides cues estimating positions joints. Following idea,...

Multi-person 3D motion prediction is an emerging task that involves predicting the future of multiple individuals based on current observations. In contrast to for a single person, this requires strong emphasis learning interacting dynamics among individuals. Broadly speaking, methods can be categorized into two groups: The first group straightforward adaptation models originally developed single-person scenarios multi-person scenarios, which evidently suboptimal. second focuses utilizing...

We made a three-channel endorectal coil for prostate imaging on the 5T magnetic resonance system and compared performance of coils with same structure 3T system. Prostate images at resolution 0.78 mm × 2.5 were performed. In vivo study shows that works better image quality surface array

Incorporating bipyridine units into the matrix of conjugated microporous polymers (CMPs) boosts exciton dissociation, charge separation, and oxidation capacity. Consequently, bipyridine-modified CMPs exhibit an impressive U(VI) separation rate (404...

Motivation: Extraocular rectus muscles (ERMs) are essential for precise eye movement control, and their assessment is vital diagnosing conditions. Magnetic Resonance Imaging (MRI) preferred due to its superior soft-tissue resolution, but it has limitations. Goal(s): To determine whether Deep Learning Reconstruction (DLR) can enhance image quality diagnosis. Approach: A study with 28 patients used DLR on ocular MRI. Two readers evaluated the images independently, statistical analyses were...

In order to reduce the noise effect in sonar images, algorithm combined morphology with median filter was proposed wipe off Gaussian and pulse simultaneously which based on simplified coupled neural networks (PCNN). Using PCNN's output images we could confirm pixel's position corresponding noise, keep internal integrity of targets by morphological filter. Then used for suppressing adjustment gray-level values iterative PCNN would noise. Simulation results show that our method's effectiveness...

This work proposes a graph search based method for human motion sequence synthesis, complementing the modern generative model (e.g., variational auto-encoder or Gaussian process) solutions that currently dominate this task and showing strong advantages at several aspects. The cornerstone of our is novel representation which we dub as graph. Each scaffolded by set realistic sequences all training data in Human3.6M benchmark). We devise scheme adds transition edges across different sequences,...