Search Results (4)

Surface electromyography (sEMG) signals are important for assessing muscle activity, neuromuscular behavior, and movement stability. sEMG signals are widely used in athlete performance monitoring and human–machine interface applications. However, existing methods have limitations in classification, accuracy and generalization across users. In this study, a real-world dataset was generated from 30 professional wrestlers using an 8-channel system based on 10 physical movements and technical elements. Nine time-domain and energy features, mean absolute value (MAV), integrated EMG (IEMG), root mean square (RMS), simple square integral (SSI), fourth power (4POW), wavelength (WL), difference absolute standard deviation (DASDV), variance (VAR), and average amplitude change (AAC), were systematically evaluated separately and in combination. Five classifiers were compared: Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), k-Nearest Neighbor (KNN), and Neural Networks (NNs). The models were evaluated for accuracy, sensitivity, specificity, positive predictive value, and F1-score. The generalization ability was analyzed through cross-subject (24/6) and cross-session validation protocols. The nine feature combinations achieved the highest classification accuracy of 97.8% with the RF algorithm. The proposed approach can serve as a practical basis for real-time muscle activity monitoring, movement classification, and rehabilitation systems. Full article

Browning is the major physiological cause of quality loss in lotus root. This study explored the effects of temperature (4 °C, 25 °C, 35 °C) or ethylene (ET) on quality, especially browning, as well as polyphenol and starch metabolism in lotus root. Low temperature (4 °C) reduced browning and color changes (L*, a*), while retaining water and vitamin C (Vc) content. ET maintained Vc and soluble protein, while high temperature (35 °C) promoted total soluble solids (TSS) and soluble sugar accumulation. ET or 35 °C upregulated polyphenol metabolism-related genes including NnPAL1/4, NnCHS1, NnF3H and NnANR, increased total phenolic and flavonoid content, and enhanced antioxidant capacity. Moreover, 35 °C increased PAL activity, and ET also upregulated NnUGT88B1. Furthermore, 4 °C downregulated NnGBE1-1/2, promoted starch accumulation, while ET upregulated NnSSI, downregulated NnGBE1-1/2, and delayed starch decline. Meanwhile, ET elevated NnETR and NnEBF1-2 and mediated ethylene signaling transduction. In conclusion, 4 °C storage was optimal for delaying browning and starch metabolism of lotus root. Meanwhile, ET treatment or 35 °C were more beneficial to obtain more phenolics and flavonoids. Full article

As eyewear devices such as smart glasses become more common, it is important to provide input methods that can be used at all times for such situations and people. Silent speech interaction (SSI) has the potential to be useful as a hands-free input method for various situations and people, including those who have difficulty with voiced speech. However, previous methods have involved sensor devices that are difficult to use anytime and anywhere. We propose a method for SSI that involves using an eyewear device equipped with infrared distance sensors. The proposed method measures facial skin movements associated with speech from the infrared distance sensor mounted on an eyewear device and recognizes silent speech commands by applying machine learning to time series sensor data. The proposed method was applied to a prototype system including a sensor device consisting of eyewear and ear-mounted microphones to measure the movements of the cheek, jaw joint, and jaw. Evaluations 1 and 2 showed that five speech commands could be recognized with an F value of 0.90 and ten longer speech commands with an F value of 0.83. Evaluation 3 showed how the recognition accuracy changes with the combination of sensor points. Evaluation 4 examined whether the proposed method can be used for a larger number of speech commands with 21 commands by using deep learning LSTM and a combination of DTW and kNN. Evaluation 5 examined the recognition accuracy in some situations affecting recognition accuracy such as re-attaching devices and walking. These results show the feasibility of the proposed method for a simple hands-free input interface, such as with media players and voice assistants. Our study provides the first wearable sensing method that can easily apply SSI functions to eyewear devices. Full article

Sketch-based image retrieval (SBIR) refers to a sub-class of content-based image retrieval problems where the input queries are ambiguous sketches and the retrieval repository is a database of natural images. In the zero-shot setup of SBIR, the query sketches are drawn from classes that do not match any of those that were used in model building. The SBIR task is extremely challenging as it is a cross-domain retrieval problem, unlike content-based image retrieval problems because sketches and images have a huge domain gap. In this work, we propose an elegant retrieval methodology, StyleGen, for generating fake candidate images that match the domain of the repository images, thus reducing the domain gap for retrieval tasks. The retrieval methodology makes use of a two-stage neural network architecture known as the stacked Siamese network, which is known to provide outstanding retrieval performance without losing the generalizability of the approach. Experimental studies on the image sketch datasets TU-Berlin Extended and Sketchy Extended, evaluated using the mean average precision (mAP) metric, demonstrate a marked performance improvement compared to the current state-of-the-art approaches in the domain. Full article