YOLSO: You Only Look Small Object

Small object detection is challenging and far from satisfactory. Most general object detectors suffer from two critical issues with small objects: (1) Feature extractor based on classification network cannot express the characteristics of small objects reasonably due to insufficient appearance information of targets and a large amount of background interference around them. (2) The detector requires a much higher location accuracy for small objects than for general objects. This paper proposes an effective and efficient small object detector YOLSO to address the above problems. For feature representation, we analyze the drawbacks in previous backbones and present a Half-Space Shortcut(HSSC) module to build a background-aware backbone. Furthermore, a coarse-to-fine Feature Pyramid Enhancement(FPE) module is introduced for layer-wise aggregation at a granular level to enhance the semantic discriminability. For loss function, we propose an exponential L1 loss to promote the convergence of regression, and a focal IOU loss to focus on prime samples with high classification confidence and high IOU. Both of them significantly improves the location accuracy of small objects. The proposed YOLSO sets state-of-the-art results on two typical small object datasets, MOCOD and VeDAI, at a speed of over 200 FPS. In the meantime, it also outperforms the baseline YOLOv3 by a wide margin on the common COCO dataset.     

Benchmark experiment of large-angle scattering reaction cross section of iron at 14 MeV using two shadow bars – Comparison of experimental results with ENDF/B-VIII –

ABSTRACT

It is important to perform neutron transport simulations with accurate nuclear data in the neutronics design of a fusion reactor. However, absolute values of large-angle scattering cross sections vary among nuclear data libraries even for well-examined nuclide of iron. Benchmark experiments focusing on large-angle scattering cross sections were thus performed to confirm the correctness of nuclear data libraries. The series benchmark experiments were performed at a DT neutron source facility, OKTAVIAN of Osaka University, Japan, by the unique experimental system established by the authors’ group, which can extract only the contribution of large-angle scattering reactions. This system consists of two shadow bars, target plate (iron), and neutron detector (niobium). Two types of shadow bars were used and four irradiations were conducted for one experiment, so that contribution of room-return neutrons was effectively removed and only large-angle scattering neutrons were extracted from the measured four Nb reaction rates. The obtained experimental results were compared with calculations for five nuclear data libraries including JENDL-4.0, JEFF.-3.3, FENDL-3.1, ENDF/B- VII, and recently released ENDF/B-VIII. It was found from the comparison that ENDF/B-VIII showed the best result, though ENDF/B-VII showed overestimation and others are in large underestimation at 14 MeV.     

Agronomic performance of high oleic,low linolenic soybean in Tennessee

Soybean oil hydrogenation alters the linolenic acid molecule to prevent the oil from becoming rancid, however, health reports have indicated trans-fat caused by hydrogenation, is not generally regarded as safe. Typical soybeans contain approximately 80 g kg⁻¹ to 120 g kg⁻¹ linolenic acid and 240 g kg⁻¹ of oleic acid. In an effort to accommodate the need for high-quality oil, the United Soybean Board introduced an industry standard for a high oleic acid greater than 750 g kg⁻¹ and linolenic acid less than 30 g kg⁻¹ oil. By combing mutations in the soybean plant at four loci, FAD2-1A and FAD2-1B, oleate desaturase genes and FAD3A and FAD3C, linoleate desaturase genes, and seed oil will not require hydrogenation to prevent oxidation and produce high-quality oil. In 2017 and 2018, a study comparing four near-isogenic lines across multiple Tennessee locations was performed to identify agronomic traits associated with mutations in FAD3A and FAD3C loci, while holding FAD2-1A and FAD2-1B constant in the mutant (high oleic) state. Soybean lines were assessed for yield and oil quality based on mutations at FAD2-1 and FAD3 loci. Variations of wild-type and mutant genotypes were compared at FAD3A and FAD3C loci. Analysis using a generalized linear mixed model in SAS 9.4, indicated no yield drag or other negative agronomic traits associated with the high oleic and low linolenic acid genotype. All four mutations of fad2-1A, fad2-1B, fad3A, and fad3C were determined as necessary to produce a soybean with the new industry standard (>750 g kg⁻¹ oleic and <30 g kg⁻¹ linolenic acid) in a maturity group-IV-Late cultivar for Tennessee growers.     

Zika Virus Growth in Human Kidney Cells Is Restricted by an Elevated Glucose Level

Mosquito-borne Zika virus (ZIKV) became a real threat to human health due to the lack of vaccine and effective antiviral treatment. The virus has recently been responsible for a global outbreak leading to millions of infected cases. ZIKV complications were highlighted in adults with Guillain–Barré syndrome and in newborns with increasing numbers of congenital disorders ranging from mild developmental delays to fatal conditions. The ability of ZIKV to establish a long-term infection in diverse organs including the kidneys has been recently documented but the consequences of such a viral infection are still debated. Our study aimed to determine whether the efficiency of ZIKV growth in kidney cells relates to glucose concentration. Human kidney HK-2 cells were infected with different ZIKV strains in presence of normal and high glucose concentrations. Virological assays showed a decrease in viral replication without modifying entry steps (viral binding, internalization, fusion) under high glucose conditions. This decrease replication was associated with a lower virus progeny and increased cell viability when compared to ZIKV-infected HK-2 cells in normal glucose concentration. In conclusion, we showed for the first time that an elevated glucose level influences ZIKV replication level with an effect on kidney cell survival.     

Patterns in environmental priorities revealed through government open data portals

The ways in which environmental priorities are framed are varied and influenced by political forces. One technological advance--the proliferation of government open data portals (ODPs)--has the potential to improve governance through facilitating access to data. Yet it is also known that the data hosted on ODPs may simply reflect the goals and interests of multiple levels of political power. In this article, I use traditional statistical correlation and regression techniques along with newer natural language processing and machine learning algorithms to analyze the corpus of datasets hosted on government ODPs (total: 49,066) to extract patterns that relate scales of governance and political liberalism/conservatism to the priorities and meaning attached to environmental issues. I find that state-level and municipal-level ODPs host different categories of environmental datasets, with municipal-level ODPs generally hosting more datasets pertaining to services and amenities and state-level ODPs hosting more datasets pertaining to resource protection and extraction. Stronger trends were observed for the influences of political conservatism/liberalism among state-level ODPs than for municipal-level ODPs.     

喷油方式对斜齿轮齿面温度的影响

为探讨喷油方式对齿面温度的影响，以斜齿轮为研究对象，通过CFD仿真和试验研究不同喷油方式对斜齿轮齿面温度的影响。针对斜齿轮喷油润滑过程的特点，分析其齿面摩擦生热量，简化轮齿喷油润滑计算模型，并基于CFX进行齿面对流换热仿真研究，得到了不同喷油方式下的齿面温度；基于功率开放式斜齿轮试验平台，对不同喷油方式下的齿面温度进行了试验测量，与仿真计算结果进行对比分析。结果表明：在高速重载斜齿轮传动系统中，啮出侧喷油润滑的方式比啮入侧喷油润滑方式的冷却降温效果要好。     

Emergency egress for the elderly in care home fire situations

Practitioners are continuing to develop egress modelling software for the design of the built environment. These models require data about human behaviour and factors for calibration, validation and verification. This study aims to address the specific data and knowledge gap: emergency egress of the elderly. Such data are difficult to collect given privacy and consent concerns, with strong relationships generally being required between residences and researchers. Through the observation of nine fire drills at six Canadian long-term care (LTC) and retirement homes, specific evacuation actions and behaviour were observed for 37 staff members and information about the evacuation of 56 residents was collected. These drills demonstrated that emergency egress in LTC and retirement homes is highly staff dependent with 72% of residents recorded requiring full assistance at all stages of movement in evacuation, and that the type of announced/unannounced drill and level of resident care will affect the type of data collected. The development of travel speed and pre-movement is discussed subject to limitation with qualitative behavioural insights of residents that were observed. This study provides valuable methodological discussion on how to conduct behavioural studies in similar highly restricted research environments. Specific attention is given to understanding the considerations that must be made when using fire drills as data sources, and the impact that these can have on using such data for modelling. This study may inform the initial setup and programming of evacuation models from an actions and behavioural perspectives of staff members and residents.     

A self-adaptive synthetic over-sampling technique for imbalanced classification

Traditionally, in supervised machine learning, (a significant) part of the available data (usually 50%-80%) is used for training and the rest—for validation. In many problems, however, the data are highly imbalanced in regard to different classes or does not have good coverage of the feasible data space which, in turn, creates problems in validation and usage phase. In this paper, we propose a technique for synthesizing feasible and likely data to help balance the classes as well as to boost the performance in terms of confusion matrix as well as overall. The idea, in a nutshell, is to synthesize data samples in close vicinity to the actual data samples specifically for the less represented (minority) classes. This has also implications to the so-called fairness of machine learning. In this paper, we propose a specific method for synthesizing data in a way to balance the classes and boost the performance, especially of the minority classes. It is generic and can be applied to different base algorithms, for example, support vector machines, k-nearest neighbour classifiers deep neural, rule-based classifiers, decision trees, and so forth. The results demonstrated that (a) a significantly more balanced (and fair) classification results can be achieved and (b) that the overall performance as well as the performance per class measured by confusion matrix can be boosted. In addition, this approach can be very valuable for the cases when the number of actual available labelled data is small which itself is one of the problems of the contemporary machine learning.