Change captioning tasks aim to detect changes in image pairs observed before and after a scene change generate natural language description of the changes. Existing studies have mainly focused on single change. However, detecting describing multiple changed parts is essential for enhancing adaptability complex scenarios. We solve above issues from three aspects: (i) propose simulation-based multi-change dataset; (ii) benchmark existing state-of-the-art methods captioning; (iii) further...

In this letter, we propose a framework that recognizes and describes changes occur in scene observed from multiple viewpoints natural language text. The ability to recognize describe occurred 3D plays an essential role variety of human-robot interaction applications. However, most current vision studies have focused on understanding the static scene. Existing change captioning approaches generate captions single-view images. Those methods limited deal with camera movement, object occlusion,...

We investigated the usefulness of probabilistic Markov chain Monte Carlo (MCMC) methods for solving magnetoencephalography (MEG) inverse problem, by using an algorithm composed combination two MCMC samplers: Reversible Jump (RJ) and Parallel Tempering (PT). The MEG problem was formulated in a Bayesian approach, we describe how RJ PT algorithms are fitted to our application. This approach offers better resolution even when number source dipoles is unknown (RJ), significant reduction...

Obtaining continuous representations of structural data such as directed acyclic graphs (DAGs) has gained attention in machine learning and artificial intelligence. However, embedding complex DAGs which both ancestors descendants nodes are exponentially increasing is difficult. Tackling this problem, we develop Disk Embeddings, a framework for into quasi-metric spaces. Existing state-of-the-art methods, Order Embeddings Hyperbolic Entailment Cones, instances Embedding Euclidean space spheres...

This study proposes a framework for describing scene change using natural language text based on indoor observations conducted before and after change. The recognition of changes plays an essential role in variety real-world applications, such as anomaly detection. Most understanding research has focused static scenes. existing captioning methods detect from single-view RGB images, neglecting the underlying three-dimensional structures. Previous use simulated scenes consisting geometry...

Although it is desirable for wheelchair users to go out alone by operating wheelchairs on their own, they are often accompanied caregivers or companions. In designing robotic wheelchairs, therefore, important consider not only how assist the user but also reduce companions' load and support activities. We specially focus communications among companions because face-to-face communication known be effective ameliorate elderly mental health. Hence, we proposed a able move alongside companion....

In this paper, we propose a robot system able to take visitors on guided tours in museum. When developing capable of giving tour an actual museum, it is necessary implement measure the location and orientation using bearing sensors installed specific environment. Furthermore, needs information pertaining both attendees other vicinity, order effectively lead for attendees. Therefore, new consisting simple elements such as laser range finder attached pole. By merely placing sensor poles, can...

This paper proposes a framework that allows the observation of scene iteratively to answer given question about scene. Conventional visual answering (VQA) methods are designed questions based on single-view images. However, in real-world applications, such as human-robot interaction (HRI), which camera angles and occluded scenes must be considered, images might difficult. Since HRI applications make it possible observe from multiple viewpoints, is reasonable discuss VQA task multi-view...

This paper introduces a system to track visitors' positions and movement patterns through museum art galleries with the goal of assisting curators other Museum Professionals (MPs) in tedious task analyzing activities, behaviors experiences. We use multiple LIDARs as primary sensing technology supporting our proposed system. also describe how valuable information related visitor can be autonomously collected analyzed using Additionally, we present solution visualize preferences respect...

Although it is desirable for wheelchair users to move alone by operating wheelchairs themselves, they are often pushed or accompanied caregivers, especially in elderly care facilities. In designing robotic wheelchairs, therefore, important consider how reduce the caregiver's load and support their activities. We particularly focus on communication among caregivers because face-to-face known be effective ameliorating mental health issues. Hence, we propose a maneuverable that can autonomously...

Recently, several robotic/intelligent wheelchairs equipped with user-friendly interfaces and/or autonomous functions to fulfill set goals have been proposed. Although wheelchair users may wish go out alone, caregivers often accompany them. Therefore, it is important consider how reduce the caregivers' load while supporting their activities and facilitating communication between caregiver user. Hence, we proposed a robotic that can move alongside by observing his/her behavior, such as body...

In order to reduce the burden of caregivers facing an increased demand for care, particularly elderly, we developed a system whereby multiple robotic wheelchairs can automatically move alongside companion. This enables small number people assist substantially larger wheelchair users effectively. utilizes environmental map and estimation position accurately identify positional relations among caregiver (or companion) each wheelchair. The are consequently able follow along even if cannot be...

PerFication is a person identification technique that uses 2D LiDAR sensor in customized dataset KoLaSu (Kobayashi Laboratory of Saitama University). Video-based recognition systems are highly effective and now at the forefront research. However, it experiences bottlenecks. New inventions can cause embarrassing situations, settings, momentum. To address limitations technology, one must introduce new technology to enhance it. Using biometric characteristics reliable valuable methods for...

We have been conducting research on ways to contribute reducing the workload of caregivers. developed a wheelchair capable following single accompanying caregiver in parallel position. However, observations actual care facilities revealed that people often move as group: for instance, two caregivers each pushed together. Therefore, we now aim develop robotic system allows multiple wheelchairs and coordinate their movement such situation necessitates complex system. Making practical...

The paper proposes a method for automatic multi-view dataset construction based on formula-driven supervised learning (FDSL). Although data collection and human annotation of 3D objects are labor-intensive, we automatically generate their training labels in the proposed dataset. To create large-scale dataset, employ fractal geometry, which is considered background information many real world. We project circle from rendered models to construct Multi-view Fractal DataBase (MV-FractalDB), then...

Daily indoor scenes often involve constant changes due to human activities. To recognize scene changes, existing change captioning methods focus on describing from two images of a scene. However, accurately perceive and appropriately evaluate physical then identify the geometry changed objects, recognizing localizing in 3D space is crucial. Therefore, we propose task explicitly localize bounding boxes point clouds describe detailed including types, object attributes, spatial locations....

Change detection is the study of detecting changes between two different images a scene taken at times. The change methodology can provide us information in which area changed time by time. However, for application use, especially on disaster investigation, it highly required to understand not only where but also what are occurred high precision and resolution. paper proposes concept semantic detection, involves intuitively inserting meaning into detected areas. We mainly focus novel...

The paper presents a novel concept that analyzes and visualizes worldwide fashion styles. Our goal is to reveal web-based viral To achieve the fashion-based analysis, we have collected culture database (FCDB), which consists of 76 million geo-tagged images in 16 cosmopolitan cities. allows us grasp trend mixed styles with descriptor codeword vector. In order unveil trends FCDB, applied simple technique temporal subtraction between consecutive vectors two different times. experiments, show...

This introduces a method which uses LIDAR to identify humans and track their positions, body orientation, movement trajectories in any public space read various types of behavioral responses surroundings. We use network poles, installed at the shoulder level typical adults reduce potential occlusion between persons and/or objects even large-scale social environments. With this arrangement, simple but effective human tracking is proposed that works by combining multiple sensors' data so areas...