Fine-grained visual classification (FGVC) which aims at recognizing objects from subcategories is a very challenging task due to the inherently subtle inter-class differences. Most existing works mainly tackle this problem by reusing the backbone network to extract features of detected discriminative regions. However, this strategy inevitably complicates the pipeline and pushes the proposed regions to contain most parts of the objects thus fails to locate the really important parts. Recently, vision transformer (ViT) shows its strong performance in the traditional classification task. The self-attention mechanism of the transformer links every patch token to the classification token. In this work, we first evaluate the effectiveness of the ViT framework in the fine-grained recognition setting. Then motivated by the strength of the attention link can be intuitively considered as an indicator of the importance of tokens, we further propose a novel Part Selection Module that can be applied to most of the transformer architectures where we integrate all raw attention weights of the transformer into an attention map for guiding the network to effectively and accurately select discriminative image patches and compute their relations. A contrastive loss is applied to enlarge the distance between feature representations of confusing classes. We name the augmented transformer-based model TransFG and demonstrate the value of it by conducting experiments on five popular fine-grained benchmarks where we achieve state-of-the-art performance. Qualitative results are presented for better understanding of our model.
Digital coding metasurface is aimed at building up a bridge between physics and information science. Higher information capacity of the digital coding metasurface means more powerful ability to control electromagnetic waves. Here, a multiband digital coding metasurface to improve the information capacity is proposed. The digital coding structures can provide 2‐bit digital states at three separate frequency bands (C, X, and Ku). It is shown that the proposed metasurface can eliminate the in‐band interference and path degradation by introducing an operator of frequency‐hopping spread spectra, from which flexible beam controls can be designed independently in every operating band. To demonstrate the capability and the compatibility, a multifunctional digital coding metasurface which can perform optical illusion, scattering reduction, and generation of orbital angular momentum in a shared aperture is presented. Numerical simulations and measured results have very good agreements, verifying excellent performance of the multiband digital coding metasurface. The proposed method opens opportunities to improve the information capacity of the digital coding metasurface and paves novel ways to multitasking systems on photonic applications.
This work demonstrates the electrochemical synthesis of nanoscale gold particles using a surfactant solution. Tetradodecylammonium bromide ͑TTAB͒ surfactant was applied to stabilize the gold clusters. Experimental results reveal that the size of the produced gold nanoparticles is controlled by the amount of TTAB surfactant, the current density, and the growth temperature. The size of the gold nanoparticles can be controlled in the range 58.3-8.3 nm. The particle size decreases as the amount of TTAB increases from 1 to 90 mg. The optimal current density in this study was 3 mA/cm 2 . The size of the produced nanoparticles increases linearly with the growth temperature from 25 to 60°C. The gold nanoparticles were observed by transmission electron microscopy, ultraviolet-visible spectrometry, and X-ray photoelectron spectroscopy. A mechanism for electrochemically controlling the size of the gold nanoparticles is presented.Nanoscale materials are of great interest due to their unique optical, electrical, and magnetic properties. Extensive investigations of gold nanoparticles in biology, nonlinear optical switching, the formation of modified surfaces for surface-enhanced Raman scattering, immunoassay labeling, optical contrast agents, and catalysis revealed that the size and shape of the particles strongly determine their physical and chemical properties. 1-6 Hence, controlling the particle size is very important. The electrochemical production of nanoparticles has been widely studied since the early work of Reetz et al. in 1994. 7,8 Their studies indicated that size-selective nanosized transition metal particles could be prepared electrochemically using tetraalkylammonium salts as stabilizers of metal clusters in a nonaqueous medium. The electrochemical method has been demonstrated to be superior to other nanoparticle production approaches because of its lower processing temperature, modest equipment, ease of controlling the yield, low cost, and high quality. 9-14 A recent study synthesized gold nanorods electrochemically by introducing a shapeinducing cosurfactant. 15 Yin et al. 16 developed a novel electrochemical technique for the size-controlled synthesis of spherical nanoparticles in poly͑N-vinylpyrrolidone͒ solution. Bartlett et al. 17 and Wiley et al. 18 reported the deposition of metal using other electrochemical approaches. The authors' research group developed the electrochemical method proposed herein to form crooked gold nanocrystals with a novel structure, using micelle templates formed from two surfactants with isopropanol addition. 19,20 In addition, carefully controlling the amount of acetone solvent added to the solution of surfactants changes the shape of the gold nanoparticles from spherical to cubic. 21 This work experimentally studies the synthesis of gold nanoparticles using a two-electrode electrochemical cell in surfactant solution, with special emphasis on the characteristics of the gold nanoparticles produced. The amount of surfactant, the current density, and the growth temperature are al...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.