Search for: All records

Corrosion is a prevalent issue in numerous industrial fields, causing expenses nearing $3 trillion or 4% of the GDP annually with safety threats and environmental pollution. To timely qualify and validate new corrosion-inhibiting materials on a large scale, accurate and efficient corrosion assessment is crucial. Yet it is hindered by a lack of automatic tools for expert-level corrosion segmentation of material science experimental images. Developing such tools is challenging due to limited domain-valid data, image artifacts visually similar to corrosion, various corrosion morphology, strong class imbalance, and millimeter-precision corrosion boundaries. To help the community address these challenges, we curate the first expert-level segmentation annotations for a real-world image dataset [1] for scientific corrosion segmentation. In addition, we design a deep learning-based model, called DeepSC-Edge that achieves guidance of ground-truth edge learning by adopting a novel loss that avoids over-fitting to edges. It also is enriched by integrating a class-balanced loss that improves segmentation with small area but crucial edges of interest for scientific corrosion assessment. Our dataset and methods pave the way to advanced deep-learning models for corrosion assessment and generation – promoting new research to connect computer vision and material science discovery. Once the appropriate approvals have been cleared, we expect to release the code and data at: https://arl.wpi.edu/ 

Corrosion of materials impacts critical economic sectors from infrastructure, transportation, defense, health, to the environment. The development of safe anti-corrosive materials is thus an important area of study in materials science. Corrosion science of preparing materials and then monitoring their corrosion under adverse conditions is labor intensive, time consuming, and extremely costly. While deep learning has become popular in automating various engineering tasks, the development of deep models for corrosion assessment is lacking. We are the first to study deep domain adaptation (DA) models for the automated assessment of the corrosion status of anti-corrosive materials. Corrosion data, i.e., photographic images of treated corroding materials, is abundant when produced in artificially controlled laboratory settings, while corrosion image data sets from rich natural outdoor environments are more challenging to produce and thus much smaller. We leverage the more readily available indoor corrosion data to train a classifier and then transfer it via deep domain adaptation to also perform well on the small yet more realistic outdoor corrosion image data set – without requiring target labels. We empirically compare 5 popular domain adaptation models on real-world corrosion image data sets. Our study finds that DA achieves 27% improvement in test accuracy compared to the performance of the no-DA baseline for classifying real-world outdoor corrosion data. 

We recently reported on the radio-frequency attenuation length of cold polar ice at Summit Station, Greenland, based on bistatic radar measurements of radio-frequency bedrock echo strengths taken during the summer of 2021. Those data also include echoes attributed to stratified impurities or dielectric discontinuities within the ice sheet (layers), which allow studies of a) estimation of the relative contribution of coherent (discrete layers, e.g.) vs. incoherent (bulk volumetric, e.g.) scattering, b) the magnitude of internal layer reflection coefficients, c) limits on the azimuthal asymmetry of reflections (birefringence), and d) limits on signal dispersion in-ice over a bandwidth of ~100 MHz. We find that i) after averaging 10000 echo triggers, reflected signal observable over the thermal floor (to depths of approximately 1500 m) are consistent with being entirely coherent, ii) internal layer reflection coefficients are measured at approximately -60 to -70 dB, iii) birefringent effects for vertically propagating signals are smaller by an order of magnitude relative to comparable studies performed at South Pole, and iv) within our experimental limits, glacial ice is non-dispersive over the frequency band relevant for neutrino detection experiments. 

Abstract. The IceCube Neutrino Observatory instruments about 1 km3 of deep, glacial ice at the geographic South Pole. It uses 5160 photomultipliers to detect Cherenkov light emitted by charged relativistic particles. An unexpected light propagation effect observed by the experiment is an anisotropic attenuation, which is aligned with the local flow direction of the ice. We examine birefringent light propagation through the polycrystalline ice microstructure as a possible explanation for this effect. The predictions of a first-principles model developed for this purpose, in particular curved light trajectories resulting from asymmetric diffusion, provide a qualitatively good match to the main features of the data. This in turn allows us to deduce ice crystal properties. Since the wavelength of the detected light is short compared to the crystal size, these crystal properties include not only the crystal orientation fabric, but also the average crystal size and shape, as a function of depth. By adding small empirical corrections to this first-principles model, a quantitatively accurate description of the optical properties of the IceCube glacial ice is obtained. In this paper, we present the experimental signature of ice optical anisotropy observed in IceCube light-emitting diode (LED) calibration data, the theory and parameterization of the birefringence effect, the fitting procedures of these parameterizations to experimental data, and the inferred crystal properties.