The low-level details and high-level semantics are both essential to the semantic segmentation task. However, to speed up the model inference, current approaches almost always sacrifice the low-level details, which leads to a considerable accuracy decrease. We propose to treat these spatial details and categorical semantics separately to achieve high accuracy and high efficiency for real-time semantic segmentation. To this end, we propose an efficient and effective architecture with a good trade-off between speed and accuracy, termed Bilateral Segmentation Network (BiSeNet V2). This architecture involves: (i) a Detail Branch, with wide channels and shallow layers to capture low-level details and generate high-resolution feature representation; (ii) a Semantic Branch, with narrow channels and deep layers to obtain high-level semantic context. The Semantic Branch is lightweight due to reducing the channel capacity and a fast-downsampling strategy. Furthermore, we design a Guided Aggregation Layer to enhance mutual connections and fuse both types of feature representation. Besides, a booster training strategy is designed to improve the segmentation performance without any extra inference cost. Extensive quantitative and qualitative evaluations demonstrate that the pro-posed architecture performs favourably against a few state-of-the-art real-time semantic segmentation approaches. Specifically, for a 2,048×1,024 input, we achieve 72.6% Mean IoU on the Cityscapes test set with a speed of 156 FPS on one NVIDIA GeForce GTX 1080 Ti card, which is significantly faster than existing methods, yet we achieve better segmentation accuracy. Code and trained models will be made publicly available.
Kolmogorov's goodness-of-fit measure, Dn, for a sample CDF has consistently been set aside for methods such as the D + n or D − n of Smirnov, primarily, it seems, because of the difficulty of computing the distribution of Dn. As far as we know, no easy way to compute that distribution has ever been provided in the 70+ years since Kolmogorov's fundamental paper. We provide one here, a C procedure that provides Pr(Dn < d) with 13-15 digit accuracy for n ranging from 2 to at least 16000. We assess the (rather slow) approach to limiting form, and because computing time can become excessive for probabilities>.999 with n's of several thousand, we provide a quick approximation that gives accuracy to the 7th digit for such cases.
Quantum fluctuations of the gravitational field in the early Universe, amplified by inflation, produce a primordial gravitational-wave background across a broad frequency band. We derive constraints on the spectrum of this gravitational radiation, and hence on theories of the early Universe, by combining experiments that cover 29 orders of magnitude in frequency. These include Planck observations of cosmic microwave background temperature and polarization power spectra and lensing, together with baryon acoustic oscillations and big bang nucleosynthesis measurements, as well as new pulsar timing array and ground-based interferometer limits. While individual experiments constrain the gravitational-wave energy density in specific frequency bands, the combination of experiments allows us to constrain cosmological parameters, including the inflationary spectral index n t and the tensor-to-scalar ratio r. Results from individual experiments include the most stringent nanohertz limit of the primordial background to date from the Parkes Pulsar Timing Array, Ω GW ðfÞ < 2.3 × 10 −10 . Observations of the cosmic microwave background alone limit the gravitational-wave spectral index at 95% confidence to n t ≲ 5 for a tensor-toscalar ratio of r ¼ 0.11. However, the combination of all the above experiments limits n t < 0.36. Future * paul.lasky@monash.eduPublished by the American Physical Society under the terms of the Creative Commons Attribution 3.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.PHYSICAL REVIEW X 6, 011035 (2016) 2160-3308=16=6(1)=011035 (11) 011035-1 Published by the American Physical Society Advanced LIGO observations are expected to further constrain n t < 0.34 by 2020. When cosmic microwave background experiments detect a nonzero r, our results will imply even more stringent constraints on n t and, hence, theories of the early Universe.
Chronic hepatitis B virus (HBV) infection is epidemiologically associated with hepatocellular carcinoma (HCC), but its role in HCC remains poorly understood due to technological limitations. In this study, we systematically characterize HBV in HCC patients. HBV sequences were enriched from 48 HCC patients using an oligo-bead-based strategy, pooled together and sequenced using the FLX-Genome-Sequencer. In the tumors, preferential integration of HBV into promoters of genes (P < 0.001) and significant enrichment of integration into chromosome 10 (P < 0.01) were observed. Integration into chromosome 10 was significantly associated with poorly differentiated tumors (P < 0.05). Notably, in the tumors, recurrent integration into the promoter of the human telomerase reverse transcriptase (TERT) gene was found to correlate with increased TERT expression. The preferred region within the HBV genome involved in integration and viral structural alteration is at the 3'-end of hepatitis B virus X protein (HBx), where viral replication/transcription initiates. Upon integration, the 3'-end of the HBx is often deleted. HBx-human chimeric transcripts, the most common type of chimeric transcripts, can be expressed as chimeric proteins. Sequence variation resulting in non-conservative amino acid substitutions are commonly observed in HBV genome. This study highlights HBV as highly mutable in HCC patients with preferential regions within the host and virus genome for HBV integration/structural alterations.
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.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.