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.     

UV curing-assisted 3D plotting of ceramic feedstock containing thermo-regulated phase-separable,photocurable vehicle for dual-scale porosity structure

We propose a novel approach for manufacturing dual-scale porosity alumina structures by UV curing-assisted 3D plotting of a specially formulated alumina feedstock using a thermo-regulated phase separable, photocurable camphene/triethylene glycol dimethacrylate (TEGDMA) vehicle. In particular, 3D plotting process was conducted at - 5 °C, and thus an alumina suspension prepared using liquid camphene/TEGDMA at room temperature could undergo phase separation, resulting in camphene crystals surrounded by walls comprised of liquid photopolymer enclosing alumina particles. To enhance the shape retention ability of extruded filaments, polystyrene (PS) polymer was used as the tackifier. The phase-separated feedrod could be extruded favorably through a nozzle and rapidly photopolymerized by UV light during the 3D plotting process. Three-dimensionally interconnected macropores were tightly constructed, which were separated by microporous alumina filaments, where micropores were created by the removal of camphene crystals via freeze-dying. The macroporosity of porous alumina ceramics was controlled by adjusting the distance between deposited filaments, while their microporosity was kept constant, leading to tightly tailored overall porosity and mechanical properties.     

Investigation of liquid water heterogeneities in large area proton exchange membrane fuel cells using a Darcy two-phase flow model in a multiphysics code

Proper management of the liquid water and heat produced in proton exchange membrane (PEM) fuel cells remains crucial to increase both its performance and durability. In this study, a two-phase flow and multicomponent model, called two-fluid model, is developed in the commercial COMSOL Multiphysics® software to investigate the liquid water heterogeneities in large area PEM fuel cells, considering the real flow fields in the bipolar plate. A macroscopic pseudo-3D multi-layers approach has been chosen and generalized Darcy's relation is used both in the membrane-electrode assembly (MEA) and in the channel. The model considers two-phase flow and gas convection and diffusion coupled with electrochemistry and water transport through the membrane. The numerical results are compared to one-fluid model results and liquid water measurements obtained by neutron imaging for several operating conditions. Finally, according to the good agreement between the two-fluid and experimentation results, the numerical water distribution is examined in each component of the cell, exhibiting very heterogeneous water thickness over the cell surface.     

High quality 3D reconstruction based on fusion of polarization imaging and binocular stereo vision

Polarization imaging can retrieve inaccurate objects’ 3D shapes with fine textures, whereas coarse but accurate depths can be provided by binocular stereo vision. To take full advantage of these two complementary techniques, we investigate a novel 3D reconstruction method based on the fusion of polarization imaging and binocular stereo vision for high quality 3D reconstruction. We first generate the polarization surface by correcting the azimuth angle errors on the basis of registered binocular depth, to solve the azimuthal ambiguity in the polarization imaging. Then we propose a joint 3D reconstruction model for depth fusion, including a data fitting term and a robust low-rank matrix factorization constraint. The former is to transfer textures from the polarization surface to the fused depth by assuming their relationship linear, whereas the latter is to utilize the low-frequency part of binocular depth to improve the accuracy of the fused depth considering the influences of missing-entries and outliers. To solve the optimization problem in the proposed model, we adopt an efficient solution based on the alternating direction method of multipliers. Extensive experiments have been conducted to demonstrate the efficiency of the proposed method in comparison with state-of-the-art methods and to exhibit its wide application prospects in 3D reconstruction.     

Fabrication of alumina ceramics with functional gradient structures by digital light processing 3D printing technology

Alumina ceramics with different unit numbers and gradient modes were prepared by digital light processing (DLP) 3D printing technology. The side length of each functional gradient structure was 10 mm, the porosity ratio was controlled to 70%, and the number of units were (1 × 1 × 1 unit) and (2 × 2 × 2 unit) respectively. The different gradient modes were named FCC, GFCC-1, GFCC-2 and GFCC-3. SEM, XRD, and other characterization methods proved that these gradient structures of alumina ceramics had only α-Al2O3 phase and good surface morphology. The mechanical properties and energy absorption properties of alumina ceramics with different functional gradient structures were studied by compression test. The results show that the gradient structure with 1 × 1 × 1 unit has better mechanical properties and energy absorption properties when the number of units is different. When the number of units is the same, GFCC-2 and GFCC-3 gradient structures have better compressive performance and energy absorption potential than FCC structures. The GFCC-2 gradient structure with 1 × 1 × 1 unit has a maximum compressive strength of 19.62 MPa and a maximum energy absorption value of 2.72 × 105 J/m3. The good performance of such functional gradient structures can provide new ideas for the design of lightweight and compressive energy absorption structures in the future.     

Fabrication of barium titanate ceramics via digital light processing 3D printing by using high refractive index monomer

A digital light processing (DLP) technology has been developed for 3D printing lead-free barium titanate (BTO) piezoelectric ceramics. By comparing the curing and rheological properties of slurries with different photosensitive monomer, a high refractive index monomer acryloyl morpholine (ACMO) was chosen, and a design and preparation method of BTO slurry with high solid content, low viscosity and high curing ability was proposed. By further selecting the printing parameters, the single-layer exposure time was reduced and the forming efficiency has been greatly improved. Sintered specimens were obtained after a nitrogen-air double-step debinding and furnace sintering process, and the BTO ceramics fabricated with 80 wt% slurry shows the highest relative density (95.32 %) and piezoelectric constant (168.1 pC/N). Furthermore, complex-structured BTO ceramics were prepared, impregnated by epoxy resin and finally assembly made into hydrophones, which has significance for the future design and manufacture of piezoelectric ceramic-based composites that used in functional devices.     

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.     

Feature selection methods for root-cause analysis among top-level product attributes

Manufacturing companies not only strive to deliver flawless products but also monitor product failures in the field to identify potential quality issues. When product failures occur, quality engineers must identify the root cause to improve any affected product and process. This root-cause analysis can be supported by feature selection methods that identify relevant product attributes, such as manufacturing dates with an increased number of product failures. In this paper, we present different methods for feature selection and evaluate their ability to identify relevant product attributes in a root-cause analysis. First, we compile a list of feature selection methods. Then, we summarize the properties of product attributes in warranty case data and discuss these properties regarding the challenges they pose for machine learning algorithms. Next, we simulate datasets of warranty cases, which emulate these product properties. Finally, we compare the feature selection methods based on these simulated datasets. In the end, the univariate filter information gain is determined to be a suitable method for a wide range of applications. The comparison based on simulated data provides a more general result than other publications, which only focus on a single use case. Due to the generic nature of the simulated datasets, the results can be applied to various root-cause analysis processes in different quality management applications and provide a guideline for readers who wish to explore machine learning methods for their analysis of quality data.