The development of portable volatile organic compound (VOC) sensors is essential for home healthcare and workplace safety because VOCs are environmental pollutants that may critically affect human health. Here, we report a compact and portable sensor platform based on a capacitive micromachined ultrasonic transducer (CMUT) array offering multiplex detection of various VOCs (toluene, acetone, ethanol, and methanol) using a single read-out system. Three CMUT resonant devices were functionalized with three different layers: (1) phenyl-selective peptide, (2) colloids of single-walled nanotubes and peptide, and (3) poly(styrene-co-allyl alcohol). As each device exhibited different sensitivities to the four VOCs, we performed principal component analysis to achieve selective detection of all four gases. For the simultaneous detection of VOCs using CMUT sensors, the changes in the resonant frequencies of three devices were monitored in real time, but using only a single oscillator through an electrically controlled relay to achieve compactness. In addition, by devising a wireless system, measurement results were transmitted to a smartphone to monitor the concentration of VOCs. We used multiple sensors to obtain a larger number of fingerprints for pattern recognition to enhance selectivity but interfaced these sensors with a single read-out circuit to minimize the footprint of the overall system. The compact CMUT-based sensor array based on a multiplex detection scheme is a promising sensor platform for portable VOC monitoring.

Chronic monitoring of bladder activity and urine volume is essential for patients suffering from urinary dysfunctions. However, due to the anatomy and dynamics of the bladder, chronic and precise monitoring of bladder activity remains a challenge. Here, we propose a new sensing mechanism that measures the bladder volume using a resistive ladder network with contact switches. Instead of measuring the impedance between the electrode continuously, the proposed sensor provides a digitized output (‘on’ or ‘off’) when the bladder volume reaches a certain threshold value. We present simple proof-of-concept sensors which compare the discrete-mode operation to the continuous-mode operation. In addition, by using multiple pairs of this contact-mode switch in a resistor ladder structure, we demonstrate monitoring of the bladder volume in four discrete steps using an idealized balloon and an ex vivo pig’s bladder. We implemented the resistive ladder network using a conductive polypyrrole/agarose hydrogel composite which exhibits a Young’s modulus comparable to that of the bladder wall. Compared to the continuous-mode operation, the proposed sensing mechanism is less susceptible to drift due to material degradation and environmental factors.

Nowadays, the welding process is essential in various manufacturing industrial fields, such as aerospace, vehicle production, and shipbuilding. The welding defects caused in the process need to be monitored as they can cause serious accidents and losses. Traditional computer vision methods in an industrial application are inefficient when the detection targets have variations in shape, scale, and color because the detection performance depends on the hand-crafted features. To overcome this limitation, deep learning models, such as the convolutional neural network (CNN), are applied to industrial defect detection. These CNN-based models trained on static images, however, have a low performance that cannot meet the industrial requirements. To deal with the challenge, bidirectional Convolutional Recurrent Reconstructive Network (bi-CRRN) is proposed for welding defect detection and localization based on welding video. Spatio-temporal data, specifically the forward and backward sequences, are considered in our bi-CRRN to get high detection performance. Moreover, an automatic defect detection equipment is developed to weld a material and monitor the welding bead simultaneously. We demonstrate that the proposed bi-CRRN outperforms the other segmentation network models in welding defect detection.

Currently, the welding process is an essential part of various industrial fields, such as shipbuilding, automobiles, aerospace, etc. Since welding defects can lead to serious adverse consequences, they must be carefully monitored. With captured welding images, welding defect localization can be formulated as a prediction of the defective area bounding box in the image, which can be solved through deep learning techniques. However, annotating the bounding box is required to train a deep learning network, which is costly and time consuming. As such, we propose a spatio-temporal deformable detection Transformer (STD-DETR), which requires only frame-level labels in the learning phase and localizes the welding defects in the inference phase. Weakly supervised learning is feasible in our task because the attention map can be obtained using a self-attention mechanism without any pixel-level or bounding box labels. STD-DETR is developed to extract not only spatial patterns but also temporal patterns with spatio-temporal attention. In addition, STD-DETR utilizes a deformable mechanism to calculate the attention to reduce computational complexity. Our proposed network is trained and evaluated on our custom welding defect video dataset captured by our manual data acquisition equipment. We demonstrate that the proposed STD-DETR outperforms other weakly supervised object localization models in welding defect localization and binary classification.

The main challenge of FSCIL is the trade-off between underfitting to a new session task and preventing forgetting the knowledge for earlier sessions. In this paper, we reveal that the angular space occupied by the features within the embedded area is relatively narrow. Consequently, after the base session training with the converged feature extractor, adding features of new classes easily overlaps with the previously occupied space of previous classes. Furthermore, in contrast to the base session classes, whose features are relatively well aggregated, the features of new session classes are dispersed over large regions. Thus, we propose the Incrementally Learned Angular Representation (ILAR) learning structure to address these issues. During the base session, ILAR attempts to increase the marginal spacing between the distributions of features for each class. In incremental sessions, modifications of features and classifiers corresponding to the previous session classes are limited to preserve the past knowledge while the features of new classes are readjusted. Furthermore, we generate additional features according to the statistics learned from the base session to reduce the variance caused by data shortage. Experiments proceeded on three popular benchmark datasets, including CIFAR100, miniImageNet, and CUB200. We demonstrate that the proposed method achieves state-of-the-art performances by effectively enhancing the new session classification ability while preserving the knowledge of the past sessions.