Benefit from the quick development of deep learning techniques, salient object detection has achieved remarkable progresses recently. However, there still exists following two major challenges that hinder its application in embedded devices, low resolution output and heavy model weight. To this end, this paper presents an accurate yet compact deep network for efficient salient object detection. More specifically, given a coarse saliency prediction in the deepest layer, we first employ residual learning to learn side-output residual features for saliency refinement, which can be achieved with very limited convolutional parameters while keep accuracy. Secondly, we further propose reverse attention to guide such side-output residual learning in a top-down manner. By erasing the current predicted salient regions from side-output features, the network can eventually explore the missing object parts and details which results in high resolution and accuracy. Experiments on six benchmark datasets demonstrate that the proposed approach compares favorably against state-of-the-art methods, and with advantages in terms of simplicity, efficiency (45 FPS) and model size (81 MB).
Wang et al. Localization of micro-/nano-robots Figure 1 Schematic illustration of the imaging techniques of micro-and nano-robots. The insets are reprinted with permission from
The Dark Energy Spectroscopic Instrument (DESI) embarked on an ambitious 5 yr survey in 2021 May to explore the nature of dark energy with spectroscopic measurements of 40 million galaxies and quasars. DESI will determine precise redshifts and employ the baryon acoustic oscillation method to measure distances from the nearby universe to beyond redshift z > 3.5, and employ redshift space distortions to measure the growth of structure and probe potential modifications to general relativity. We describe the significant instrumentation we developed to conduct the DESI survey. This includes: a wide-field, 3.°2 diameter prime-focus corrector; a focal plane system with 5020 fiber positioners on the 0.812 m diameter, aspheric focal surface; 10 continuous, high-efficiency fiber cable bundles that connect the focal plane to the spectrographs; and 10 identical spectrographs. Each spectrograph employs a pair of dichroics to split the light into three channels that together record the light from 360–980 nm with a spectral resolution that ranges from 2000–5000. We describe the science requirements, their connection to the technical requirements, the management of the project, and interfaces between subsystems. DESI was installed at the 4 m Mayall Telescope at Kitt Peak National Observatory and has achieved all of its performance goals. Some performance highlights include an rms positioner accuracy of better than 0.″1 and a median signal-to-noise ratio of 7 of the [O ii] doublet at 8 × 10−17 erg s−1 cm−2 in 1000 s for galaxies at z = 1.4–1.6. We conclude with additional highlights from the on-sky validation and commissioning, key successes, and lessons learned.
The triboluminescence phenomenon has been proposed as a sensor system for detecting and monitoring damage in aerospace and civil infrastructure systems (CIS). While significant work is being done in developing such systems for aerospace structures, little or no work is being done in developing triboluminescence-based sensor systems for the critical and aging CIS. This article reports some findings in the work being done by the authors to develop such a sensor system for civil infrastructure applications. A ZnS:Mn-based cementitious patch that emits light when stressed or fractured was developed and its triboluminescence performance under loading characterized. The results show that a ZnS:Mn concentration level of 10% gives the best triboluminescence response without adversely affecting the compressive strength of the patch, while also minimizing the use of the expensive ZnS:Mn crystals. The triboluminescence response increased as the concentration of ZnS:Mn in the system increased. The highest triboluminescence response was obtained at a concentration level of 25% but resulted in significant reduction in the system's compressive strength. Nonetheless, the presence of ZnS:Mn affects the hydration process by slowing down the conversion of the needleshaped crystals of calcium sulfoaluminate hydrate (ettringites) into the monosulfate hydrate that makes concrete vulnerable to sulfate attack.
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.