Naoya Yoshimura 교수

This study presents a new neural network model for recognizing manual works using body-worn accelerometers in industrial settings, named Lightweight Ordered-work Segmentation Network (LOS-Net). In industrial domains, a human worker typically repetitively performs a set of predefined processes, with each process consisting of a sequence of activities in a predefined order. State-of-the-art activity recognition models, such as encoder-decoder models, have numerous trainable parameters, making thei

Unlike human daily activities, existing publicly available sensor datasets for work activity recognition in industrial domains are limited by difficulties in collecting realistic data as close collaboration with industrial sites is required. This also limits research on and development of methods for industrial applications. To address these challenges and contribute to research on machine recognition of work activities in industrial domains, in this study, we introduce a new large-scale dataset

OpenPack is an open-access logistics dataset for human activity recognition, which contains human movement and package information from 16 subjects in four scenarios. Human movement information is subdivided into three types of data, acceleration, physiological, and depth-sensing. The package information includes the size and number of items included in each packaging job. In the "Humanware laboratory" at IST Osaka University, with the supervision of industrial engineers, an experiment to mimic

Owing to the growing demand for wearable context-aware applications, activity recognition technologies have attracted great attention. A neural network has been recently used as a recognition algorithm because of its discrimination and feature extraction ability. While understanding the network provides us useful information to improve its performance, visualization techniques for neural networks have been not explored yet in the human activity recognition field. We propose a visualization metho

Although deep learning-based activity recognition using wearable sensors has been actively studied to implement smart applications such as supporting elderly care, healthcare, and home automation, techniques for understanding the inside of activity recognition networks have not yet been investigated thoroughly. In the computer vision research field, activation maximization (AM) was proposed to visualize the internal functions of networks. However, when conventional AM techniques, which are tailo

Since ancient times, a hug has been one of the most basic ways to express emotions and has played an important role in building relationships between people. On the other hand, social robots that are designed to provide mental health care to patients have been attracting great attention, and hugging between humans and robots is becoming more and more popular. In this study, we propose a huggable robot that allows intimate interactions between humans and robots. Our robot is based on a tensegrity

Speech is a direct and intuitive method to control a robot. While natural speech can capture a rich variety of commands, verbal input is poorly suited to finer grained and real-time control of continuous actions such as short and precise motion commands. For these types of operations, continuous non-verbal speech is more suitable, but it lacks the naturalness and vocabulary breadth of verbal speech. In this work, we propose to combine the two types of vocal input by extending the last vowel of a

Inertial sensor data collected from wearable smart devices such as smartwatches are expected to be used in various smart applications such as video game controllers, hand drawing, hand writing, gestural input devices, human activity recognition, and remote communication using sign language. However, since the maximum sampling rate of inertial sensors in commercial smartwatches is restricted, capturing fine-grained body movements using the low-sampled signals is difficult for these sensors. There

This study investigates the feasibility of fine-grained gesture recognition using upsampled acceleration sensor data. Because the maximum sampling rate of smartwatch devices is limited by operating systems, we simulate high resolution acceleration data using a neural network from low resolution signals in order to capture distinguishing features of gestures containing high frequency components.

<strong>OpenPack</strong> is an open access logistics-dataset for human activity recognition, which contains human movement and package information from ten subjects. The package information includes the size and number of items included in each packaging task. While the human movement information is subdivided into three types of data, acceleration, physiological, and depth-sensing. In the "Humanware laboratory" at IST Osaka University, with the supervision of industrial engineers, an experimen

OpenPack Dataset is a new large-scale multi-modal dataset of packing processes. This artifact guide provides an overview of material useful for the OpenPack dataset users and illustrates the minimum steps required to download the dataset and load it associated with the ground truth labels.

<strong>OpenPack</strong> is an open access logistics-dataset for human activity recognition, which contains human movement and package information from 10 experienced subjects in two scenarios. The package information includes the size and number of items included in each packaging job. Human movement information is subdivided into three types of data, acceleration, physiological, and depth-sensing. In the "Humanware laboratory" at IST Osaka University, with the supervision of industrial engine

<strong>OpenPack</strong> is an open access logistics-dataset for human activity recognition, which contains human movement and package information from ten subjects. The package information includes the size and number of items included in each packaging task. While the human movement information is subdivided into three types of data, acceleration, physiological, and depth-sensing. In the "Humanware laboratory" at IST Osaka University, with the supervision of industrial engineers, an experimen

<strong>OpenPack</strong> is an open access logistics-dataset for human activity recognition, which contains human movement and package information from ten subjects. The package information includes the size and number of items included in each packaging task. While the human movement information is subdivided into three types of data, acceleration, physiological, and depth-sensing. In the "Humanware laboratory" at IST Osaka University, with the supervision of industrial engineers, an experimen

OpenPack is an open access logistics-dataset for human activity recognition, which contains human movement and package information from 10 subjects in four scenarios. Human movement information is subdivided into three types of data, acceleration, physiological, and depth-sensing. The package information includes the size and number of items included in each packaging job. In the "Humanware laboratory" at IST Osaka University, with the supervision of industrial engineers, an experiment to mimic