In this work we propose a workflow to deal with overlaid images—images with superimposed text and company logos—, which is very common in underwater monitoring videos and surveillance camera footage. It is demonstrated that it is possible to use Explaining Artificial Intelligence to improve deep learning models performance for image classification tasks in general. A deep learning model trained to classify metal surface defect, which previously had a low performance, is then evaluated with Layer-wise relevance propagation—an Explaining Artificial Intelligence technique—to identify problems in a dataset that hinder the training of deep learning models in a wide range of applications. Thereafter, it is possible to remove this unwanted information from the dataset—using different approaches: from cutting part of the images to training a Generative Inpainting neural network model—and retrain the model with the new preprocessed images. This proposed methodology improved F1 score in 20% when compared to the original trained dataset, validating the proposed workflow.

This study aimed to systematically understand the magnetic properties of magnetite (Fe₃O₄) nanoparticles functionalized with different Pluronic F-127 surfactant concentrations (Fe₃O₄@Pluronic F-127) obtained by using an improved magnetic characterization method based on three-dimensional magnetic maps generated by scanning magnetic microscopy. Additionally, these Fe₃O₄ and Fe₃O₄@Pluronic F-127 nanoparticles, as promising systems for biomedical applications, were prepared by a wet chemical reaction. The magnetization curve was obtained through these three-dimensional maps, confirming that both Fe₃O₄ and Fe₃O₄@Pluronic F-127 nanoparticles have a superparamagnetic behavior. The as-prepared samples, stored at approximately 20 °C, showed no change in the magnetization curve even months after their generation, resulting in no nanoparticles free from oxidation, as Raman measurements have confirmed. Furthermore, by applying this magnetic technique, it was possible to estimate that the nanoparticles’ magnetic core diameter was about 5 nm. Our results were confirmed by comparison with other techniques, namely as transmission electron microscopy imaging and diffraction together with Raman spectroscopy. Finally, these results, in addition to validating scanning magnetic microscopy, also highlight its potential for a detailed magnetic characterization of nanoparticles.

We improved a magnetic scanning microscope for measuring the magnetic properties of minerals in thin sections of geological samples at submillimeter scales. The microscope is comprised of a 200 µm diameter Hall sensor that is located at a distance of 142 µm from the sample; an electromagnet capable of applying up to 500 mT DC magnetic fields to the sample over a 40 mm diameter region; a second Hall sensor arranged in a gradiometric configuration to cancel the background signal applied by the electromagnet and reduce the overall noise in the system; a custom-designed electronics system to bias the sensors and allow adjustments to the background signal cancelation; and a scanning XY stage with micrometer resolution. Our system achieves a spatial resolution of 200 µm with a noise at 6.0 Hz of 300 nT_rms/(Hz)^1/2 in an unshielded environment. The magnetic moment sensitivity is 1.3 × 10⁻¹¹ Am². We successfully measured the representative magnetization of a geological sample using an alternative model that takes the sample geometry into account and identified different micrometric characteristics in the sample slice.

Scanning magnetic microscopy is a tool that has been used to map magnetic fields with good spatial resolution and field sensitivity. This technology has great advantages over other instruments; for example, its operation does not require cryogenic technology, which reduces its operational cost and complexity. Here, we presented a spatial domain technique based on an equivalent layer approach for processing the data set produced by magnetic microscopy. This approach estimated a magnetic moment distribution over a fictitious layer composed by a set of dipoles located below the observation plane. For this purpose, we formulated a linear inverse problem for calculating the magnetic vector and its amplitude. Vector field maps are valuable tools for the magnetic interpretation of samples with a high spatial variability of magnetization. These maps could provide comprehensive information regarding the spatial distribution of magnetic carriers. In addition, this approach might be useful for characterizing isolated areas over samples or investigating the spatial magnetization distribution of bulk samples at the micro and millimeter scales. This technique could be useful for many applications that require samples that need to be mapped without a magnetic field at room temperature, including rock magnetism.