Explore the vast range of titles available.

Visual analysis of complex fish habitats is an important step towards sustainable fisheries for human consumption and environmental protection. Deep Learning methods have shown great promise for scene analysis when trained on large-scale datasets. However, current datasets for fish analysis tend to focus on the classification task within constrained, plain environments which do not capture the complexity of underwater fish habitats. To address this limitation, we present DeepFish as a benchmark suite with a large-scale dataset to train and test methods for several computer vision tasks. The dataset consists of approximately 40 thousand images collected underwater from 20 habitats in the marine-environments of tropical Australia. The dataset originally contained only classification labels. Thus, we collected point-level and segmentation labels to have a more comprehensive fish analysis benchmark. These labels enable models to learn to automatically monitor fish count, identify their locations, and estimate their sizes. Our experiments provide an in-depth analysis of the dataset characteristics, and the performance evaluation of several state-of-the-art approaches based on our benchmark. Although models pre-trained on ImageNet have successfully performed on this benchmark, there is still room for improvement. Therefore, this benchmark serves as a testbed to motivate further development in this challenging domain of underwater computer vision.

Recent developments in Deep Learning have opened the possibility for automated segmentation of large and highly detailed CT scan datasets of fossil material. However, previous methodologies have required large amounts of training data to reliably extract complex skeletal structures. Here we present a method for automated Deep Learning segmentation to obtain high-fidelity 3D models of fossils digitally extracted from the surrounding rock, training the model with less than 1%-2% of the total CT dataset. This workflow has the capacity to revolutionise the use of Deep Learning to significantly reduce the processing time of such data and boost the availability of segmented CT-scanned fossil material for future research outputs. Our final Unet segmentation model achieved a validation Dice similarity of 0.96.

Surgical sterilization or neutering of dogs is a commonly performed procedure in veterinary practices in many countries. In recent decades, concerns have been raised regarding possible side effects of neutering, including increased risk of certain neoplastic, musculoskeletal and endocrinological conditions. Considering that age serves as a significant confounding factor for some of these conditions, evaluating longevity statistics could provide valuable insights into the impact of neutering. The aim of this study was to compare longevity between neutered and sexually intact male and female Rottweilers, using electronic patient records collected by the VetCompass Australia database. Male and female Rottweilers neutered before 1 year of age ( n = 207) demonstrated an expected lifespan 1.5 years and 1 year shorter, respectively, than their intact counterparts ( n = 3085; p < 0.05). Broadening this analysis to include animals neutered before the age of 4.5 years ( n = 357) produced similar results.

Subsurface injection of fluids inside tectonically active areas is in some cases followed by a seismic response. This paper examines an example of seismic activation on the northeast coast of Sakhalin Island (in the area of the North American tectonic plate) in 2013–2014, which coincides with the start of the pilot operation of injection wells designated for the disposal of drilling fluids and co-produced and domestic wastewater from nearby oil and gas operations. Hypothesizing that the recorded seismicity is entirely induced by the fluid disposal, and considering injection rates of 105 m3/year, we estimate a seismogenic index of Σ =  −3.98 ± 0.06. Assuming a total injection volume of VC = 107 m3 until 2041, the probability of occurrence of at least one earthquake with a magnitude M 5.5 was calculated to be PE = 0.99. The obtained results are of practical interest for developing seismic risk management strategies and using numerical tools to quantify seismic hazards.

Key messageWe present a comprehensive survey of cytogenetic and genomic diversity of the GGAtAt genepool of wheat, thereby unlocking these plant genetic resources for wheat improvement.Wheat yields are stagnating around the world and new sources of genes for resistance or tolerances to abiotic traits are required. In this context, the tetraploid wheat wild relatives are among the key candidates for wheat improvement. Despite its potential huge value for wheat breeding, the tetraploid GGAtAt genepool is largely neglected. Understanding the population structure, native distribution range, intraspecific variation of the entire tetraploid GGAtAt genepool and its domestication history would further its use for wheat improvement. The paper provides the first comprehensive survey of genomic and cytogenetic diversity sampling the full breadth and depth of the tetraploid GGAtAt genepool. According to the results obtained, the extant GGAtAt genepool consists of three distinct lineages. We provide detailed insights into the cytogenetic composition of GGAtAt wheats, revealed group- and population-specific markers and show that chromosomal rearrangements play an important role in intraspecific diversity of T. araraticum. The origin and domestication history of the GGAtAt lineages is discussed in the context of state-of-the-art archaeobotanical finds. We shed new light on the complex evolutionary history of the GGAtAt wheat genepool and provide the basis for an increased use of the GGAtAt wheat genepool for wheat improvement. The findings have implications for our understanding of the origins of agriculture in southwest Asia.

Logit analysis is widely used for binary data classification in geoscience. In this study, logistic regression was used as a tool for deriving the binary classifier of tsunamigenic and non-tsunamigenic earthquakes for near-source early warning. The catalogue of submarine earthquakes and tsunami database were merged into one seismic database, in which an additional binary variable associated with the tsunamigenic class (true or false) was assigned to each seismic event. The training dataset consisted of 712 M6.0+ submarine earthquakes, including 80 tsunamigenic and 632 non-tsunamigenic events that occurred in the northwestern part of the Pacific Ocean from 1960 to 2020. The target area has already experienced significant and catastrophic tsunamis. The best performance metrics were archived with the predictors given by the earthquake magnitude, logarithm of the source depth and the seafloor depth in the epicenter location. The current analysis clearly showed that the data-driven logit model significantly improved the performance metrics of the threshold magnitude criteria that are widely used by tsunami warning agencies. Authors suggested that logit-based binary classifier led to improve the tsunami alert efficiency in the Northwestern Pacific Ocean.

The paper is devoted to an original method of time-domain fluorescence molecular lifetime tomography (FMLT) based on asymptotic approximation to the fluorescence source function. Such an approximation helps to simplify the expressions that describe the FMLT reconstruction model in the time domain and to formulate the linear inverse problem for a generalized fluorescence parameter distribution function. The method firstly solves this problem and then separates distributions of the fluorophore absorption coefficient and the fluorescence lifetime from the generalized function. The paper analyzes results the authors have obtained during last 5 years in their testing the method in numerical and physical experiments. The method is inferred to be quite promising and directions of further research for its verification as a sub-millimeter resolution method are outlined.

Background Bayesian clustering algorithms, in particular those utilizing Dirichlet Processes (DP), return a sample of the posterior distribution of partitions of a set. However, in many applied cases a single clustering solution is desired, requiring a ’best’ partition to be created from the posterior sample. It is an open research question which solution should be recommended in which situation. However, one such candidate is the sample mean, defined as the clustering with minimal squared distance to all partitions in the posterior sample, weighted by their probability. In this article, we review an algorithm that approximates this sample mean by using the Hungarian Method to compute the distance between partitions. This algorithm leaves room for further processing acceleration. Results We highlight a faster variant of the partition distance reduction that leads to a runtime complexity that is up to two orders of magnitude lower than the standard variant. We suggest two further improvements: The first is deterministic and based on an adapted dynamical version of the Hungarian Algorithm, which achieves another runtime decrease of at least one order of magnitude. The second improvement is theoretical and uses Monte Carlo techniques and the dynamic matrix inverse. Thereby we further reduce the runtime complexity by nearly the square root of one order of magnitude. Conclusions Overall this results in a new mean partition algorithm with an acceleration factor reaching beyond that of the present algorithm by the size of the partitions. The new algorithm is implemented in Java and available on GitHub (Glassen, Mean Partition, 2018).

Computational models of representative volumes of metal matrix composites are crucial for investigating material behavior on the microscale. Numerical simulations often employ finite element models of representative volumes. Such models are based on observations of material microstructural properties, for example, by means of electron microscopy. Constructing a geometrical model of a representative volume for further computations can be a tedious task. This paper presents a new approach to creating geometrical models of this kind. The approach is based on the simulation of rigid body motion in a gravitational field and allows one to automatically generate geometrical models of representative volumes. The approach capabilities are exemplified by two geometrical models of representative volume. One model describes a metal matrix composite with high volume fraction of prismatic reinforcement particles, produced by liquid metal infiltration. The other model describes a metal matrix composite with high volume fraction of metal pellets, produced by powder metallurgy.

Using the development of a technique for the non-destructive testing of a fireclay product (“asterisks”) using the natural vibrations of the object under consideration as an example, this article presents a mathematical model that enables us to describe specifically the process of converting an external dynamic effect into the natural vibrations of the part. The model stands out because firstly, it is a carrier of the free energy of elastic deformation, and the natural oscillations are formed inside it. Secondly, the model represents a Riemannian space where all dynamic parameters are constant and reduced to zero, i.e., the model does not exist in the physical world, but only as a functional space. The proposed model can be used as an effective tool for analyzing the processes registered during non-destructive testing and vibration diagnostics.