Identification and morphological analysis of microorganisms are of high interest in scientific research, especially in the medical field and food industry. Identification allows rapid functional characterization based on similarities with known related species enabling to confirm the identity of an isolate used, for example, in a trademarked industrial process. Monitoring of microorganisms within a given ecosystem and analysis of the morphological characteristics of the observed species enable quality control of the process under analysis. Such procedures are carried out manually in specialized laboratories by trained personnel using the appropriate optical equipment; therefore, it may be of great interest to use automatic measurement approaches that enable rapid and effective process analysis. Artificial intelligence techniques in computer vision and especially deep learning are well suited for this purpose. This article describes the realization of an automatic measurement system based on deep learning for the identification and measurement of morphological parameters of Saccharomyces cerevisiae microorganisms present in brewer's yeast by returning for each of the objects identified within the image the confidence score, the coordinates, and the dimensions of the corresponding ellipsoid-shaped cell. The metrological characteristics of the system have been defined through a calibration process by comparing measurements with a reference system.

Vibrations in road vehicles can have highly harmful effects on both the vehicle components and the passengers. These vibrations are mainly caused by lowquality pavement, so it is important to keep the road network in good condition and to know its general quality. In our study, we present the development of a universally applicable, low-cost measurement system for the purpose of measuring the condition of pavement surfaces. The system can be used to identify road sections in urgent and near future need of maintenance, thus helping to schedule construction works efficiently. The system is based on an inertial sensor unit mounted on the vehicle suspension, in contrast to previous systems, and therefore offers an improvement in the accuracy of the measurement. In our study, a principal component analysis and time series segmentationbased algorithm is introduced to extract relevant features from the raw sensor data. Subsequently, each segment is classified into pre-defined classes based on its surface quality using a binary decision tree-based classification model fitted by supervised learning. After validation, the system is tested on public roads under real measurement conditions.

The development and spread of Machine Learning methodologies have also involved the field of anomaly detection, particularly focused on fault detection. This is one of the main goals of Industry 4.0, as it is necessary to optimize repair time and cost. In this regard, Machine Learning is needed to identify precursor features of possible failures that would be difficult for a human operator to discern. However,compared to Deep Learning methodologies, these cannot be fully automatic because of the need to make choices about the features identified by the system. This paper aims to propose a Machine Learning-based system for detecting electrical anomalies attributable to malfunctions in connected industrial machinery. Specifically, the proposal is a fusion of unsupervised learning and traditional methodology to minimize human intervention while maintaining an explainable, white-box approach, contrary to proposals based on Deep Learning. The results demonstrated better performance to techniques of the state of the art.

Flux is a necessity in the lead-free soldering process. Inactivated flux residues can cause electrical shortages and functional issues by ion migration. Because of miniaturization and the complexity of recent electronic products a cleaner manufacturing process is required. Atmospheric pressure plasma treatment (PT) is a commonly applied surface cleaning method in the manufacturing industry and has been proved to be effective in improving the wettability in case of metal and polymer surfaces. Three different types of printed circuit board (PCB) surface finishes were investigated before and after plasma treatment. The aim of this study is to investigate the mechanisms of plasma cleaning and to possibly determine the root cause of the improvement of solderability on printed circuit boards resulting from atmospheric pressure plasma treatment. For this investigation of the PCB surface finishes, scanning electron microscopy (SEM) images were taken, and the composition of the surfaces were analyzed by energy dispersive X-ray spectrometry (EDAX) and laser-induced breakdown spectrometry (LIBS) as well.

This paper proposes a multi-task learning system for industrial defect detection, focusing on surface scratches on titanium gaskets as the detection target. Our model is built using a deep convolutional neural network (CNN). As defects are infrequent in actual production lines, we adopt the Faster-RCNN object detection network architecture and integrate it with a multi-task learning system. A classification task model is introduced into the backbone network to filter out a significant number of defect-free images. This step reduces the computation and data transmission time of the subsequent target detection task, accelerating the detection process. Experimental results show that our proposed method reduces inference time by 37.6% and 42.46% in actual production lines with defect rates of 12% and 5%, respectively, while maintaining 96% of the original model's performance.