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.     

Functional Groups Assisted Tunable Dielectric Permittivity of Guest-Free Zn-Based Coordination Polymers for Gate Dielectrics

The dielectric properties of coordination polymers has been a topic of recent interest, but the role of different functional groups on the dielectric properties of these polymers has not yet been fully addressed. Herein, the effects of electron-donating (R=NH₂) and electron-withdrawing (R=NO₂) groups on the dielectric behavior of such materials were investigated for two thermally stable and guest-free Zn-based coordination polymers, [Zn(L₁)(L₂)]_n ( 1 ) and [Zn(L₁)(L₃)]_n ( 2 ) [L₁=2-(2-pyridyl) benzimidazole (Pbim), L₂=5-aminoisophthalate (Aip), and L₃=5-nitroisophthalate (Nip)]. The results of dielectric studies of 1 revealed that it possesses a high dielectric constant (κ=65.5 at 1 kHz), while compound 2 displayed an even higher dielectric constant (κ=110.3 at 1 kHz). The electron donating and withdrawing effects of the NH₂ and NO₂ substituents induce changes in the polarity of the polymers, which is due to the inductive effect from the aryl ring for both NO₂ and NH₂. Theoretical results from density functional theory (DFT) calculations, which also support the experimental findings, show that both compounds have a distinct electronic behavior with diverse wide bandgaps. The significance of the current work is to provide information about the structure-dielectric property relationships. So, this study promises to pave the way for further research on the effects of different functional groups on coordination polymers on their dielectric properties.     

High-fidelity Recognition of Organotrifluoroborate Anions (R−BF3−) as Designer Guest Molecules

The recognition of boron compounds is well developed as boronic acids but untapped as organotrifluoroborate anions (R−BF₃⁻). We are exploring the development of these and other designer anions as anion-recognition motifs by considering them as substituted versions of the parent inorganic ion. To this end, we demonstrate strong and reliable binding of organic trifluoroborates, R−BF₃⁻, by cyanostar macrocycles that are size-complementary to the inorganic BF₄⁻ progenitors. We find that recognition is modulated by the substituent's sterics and that the affinities are retained using the common K⁺ salts of R−BF₃⁻ anions.     

No-Reference Video Quality Assessment using novel hybrid features and two-stage hybrid regression for score level fusion

This paper presents a novel No-Reference Video Quality Assessment (NR-VQA) model that utilizes proposed 3D steerable wavelet transform-based Natural Video Statistics (NVS) features as well as human perceptual features. Additionally, we proposed a novel two-stage regression scheme that significantly improves the overall performance of quality estimation. In the first stage, transform-based NVS and human perceptual features are separately passed through the proposed hybrid regression scheme: Support Vector Regression (SVR) followed by Polynomial curve fitting. The two visual quality scores predicted from the first stage are then used as features for the similar second stage. This predicts the final quality scores of distorted videos by achieving score level fusion. Extensive experiments were conducted using five authentic and four synthetic distortion databases. Experimental results demonstrate that the proposed method outperforms other published state-of-the-art benchmark methods on synthetic distortion databases and is among the top performers on authentic distortion databases. The source code is available at https://github.com/anishVNIT/two-stage-vqa.     

A signal adaptive diffusion filter for video coding: Mathematical framework and complexity reductions

In this paper we combine video compression and modern image processing methods. We construct novel iterative filter methods for prediction signals based on Partial Differential Equation (PDE) based methods. The mathematical framework of the employed diffusion filter class is given and some desirable properties are stated. In particular, two types of diffusion filters are constructed: a uniform diffusion filter using a fixed filter mask and a signal adaptive diffusion filter that incorporates the structures of the underlying prediction signal. The latter has the advantage of not attenuating existing edges while the uniform filter is less complex. The filters are embedded into a software based on HEVC with additional QTBT (Quadtree plus Binary Tree) and MTT (Multi-Type-Tree) block structure. In this setting, several measures to reduce the coding complexity of the tool are introduced, discussed and tested thoroughly. The coding complexity is reduced by up to 70% while maintaining over 80% of the gain. Overall, the diffusion filter method achieves average bitrate savings of 2.27% for Random Access having an average encoder runtime complexity of 119% and 117% decoder runtime complexity. For individual test sequences, results of 7.36% for Random Access are accomplished.     

Interlayer Engineering of Molybdenum Trioxide toward High‐Capacity and Stable Sodium Ion Half/Full Batteries

Orthorhombic molybdenum trioxide (MoO₃) is one of the most promising anode materials for sodium‐ion batteries because of its rich chemistry associated with multiple valence states and intriguing layered structure. However, MoO₃ still suffers from the low rate capability and poor cycle induced by pulverization during de/sodiation. An ingenious two‐step synthesis strategy to fine tune the layer structure of MoO₃ targeting stable and fast sodium ionic diffusion channels is reported here. By integrating partially reduction and organic molecule intercalation methodologies, the interlayer spacing of MoO₃ is remarkably enlarged to 10.40 Å and the layer structural integration are reinforced by dimercapto groups of bismuththiol molecules. Comprehensive characterizations and density functional theory calculations prove that the intercalated bismuththiol (DMcT) molecules substantially enhanced electronic conductivity and effectively shield the electrostatic interaction between Na⁺ and the MoO₃ host by conjugated double bond, resulting in improved Na⁺ insertion/extraction kinetics. Benefiting from these features, the newly devised layered MoO₃ electrode achieves excellent long‐term cycling stability and outstanding rate performance. These achievements are of vital significance for the preparation of sodium‐ion battery anode materials with high‐rate capability and long cycling life using intercalation chemistry.     

Combustion properties of H2/N2/O2/steam mixtures

The present work reports new experimental and numerical results of the combustion properties of hydrogen based mixtures diluted by nitrogen and steam. Spherical expanding flames have been studied in a spherical bomb over a large domain of equivalence ratios, initial temperatures and dilutions at an initial pressure of 100 kPa (T_ini = 296, 363, 413 K; N₂/O₂ = 3.76, 5.67, 9; %Steam = 0, 20, 30). From these experiments, the laminar flame speed $S_L^0$, the Markstein length L’, the activation energy Ea and the Zel'dovich β number have been determined. These parameters were also simulated using COSILAB^® in order to verify the validity of the Mével et al. [1] detailed kinetic mechanism. Other parameters as the laminar flame thickness δ and the effective Lewis number Le_eff were also simulated. These new results aim at providing an extended database that will be very useful in the hydrogen combustion hazard assessment for nuclear reactor power plant new design.     

Yield enhancement of elliptical high harmonics driven by bicircular laser pulses

We theoretically investigate the yield enhancement of elliptical high harmonics in the interaction of molecules with bicircular laser pulses by solving the time-dependent Schrödinger equation. It is shown that by adjusting the relative intensity ratio of the two bicircular field components in specific ranges the yield of the molecular high harmonics for the plateau and cutoff regions can be respectively enhanced. To analyze this enhancement phenomenon, we calculate the weights of the electron classical trajectories. Additionally, we also study the ellipticity distribution of harmonics for different intensity ratios. We find that these enhanced harmonics are elliptically polarized, which we mainly attribute to the recombination dipole moment of the major weighted trajectories. These enhanced elliptical extreme ultraviolet and soft x-ray radiations may serve as essential tools for exploring the ultrafast dynamics in magnetic materials and chiral media.     

A single-phase algorithm for mining high utility itemsets using compressed tree structures

Mining high utility itemsets (HUIs) from transaction databases considers such factors as the unit profit and quantity of purchased items. Two-phase tree-based algorithms transform a database into compressed tree structures and generate candidate patterns through a recursive pattern-growth procedure. This procedure requires a lot of memory and time to construct conditional pattern trees. To address this issue, this study employs two compressed tree structures, namely, Utility Count Tree and String Utility Tree, to enumerate valid patterns and thus promote fast utility computation. Furthermore, the study presents an algorithm called single-phase utility computation (SPUC) that leverages these two tree structures to mine HUIs in a single phase by incorporating novel pruning strategies. Experiments conducted on both real and synthetic datasets demonstrate the superior performance of SPUC compared with IHUP, UP-Growth, and UP-Growth+ algorithms.     

The Szegö kernel on a class of non-compact CR manifolds of high codimension

We generalize Nagel’s formula for the Szegö kernel and use it to compute the Szegö kernel on a class of non-compact CR manifolds whose tangent space decomposes into one complex direction and several totally real directions. We also discuss the control metric on these manifolds and relate it to the size of the Szegö kernel.