This research explores the intricate relationship between emojis in Twitter interactions and sentiment interpretation in Japan. Emojis, as part of the digital language lexicon, have become essential in social media communication, especially on platforms like Twitter. However, integrating emojis into sentiment analysis has yet to be heavily explored, presenting an opportunity to better understand localized digital sentiment expression. This paper analyzes a data set within the COVID-19 Era of Twitter messages from Japan, focusing on emojis and the sentiments they convey. An established model was utilized to process Twitter text data and estimate their sentiment analysis, reflecting a more accurate portrayal of users' emotional states. We find a compelling link between emoji usage and the sentiment expressed. Emojis were shown to provide critical sentiment indicators often absent or ambiguous in the text alone. Certain emojis were identified as having a consistent correlation with particular sentiments.

We propose a method to protect ownership of specified areas on image data using blockchain. Different values are assigned corresponding to the object's importance in the region and the ownership of the area is managed and given to the required user. The detected areas are individually encrypted by XOR cipher, and a corresponding key image is created to decrypt to protect ownership of selected object regions detected by the object recognition algorithm. We use non-fungible tokens(NFTs) to secure key images by managing ownership of each object on image data. Key images to decrypt are registered as NFTs. Ownership NFTs are also created to access and retrieve the key image NFTs. Each key image NFT is generated by key holders and ownership NFTs are obtained by users requiring them. In addition, there is a judging function that clarifies if ownership and key NFTs match. Key NFTs appear on the screen only when they are matched.

Omnidirectional cameras, which can capture an image that covers a 360-degree view, have been widely used in recent years. This study proposes a method to detect human queues in images captured with omnidirectional cameras. The method can be applied to further research in human flow analysis and congestion estimation. In this study, we analyzed the real space using panoramic images captured by an omnidirectional camera. These images contain distortions because the spherical image captured is converted into a flat image, making feature detection difficult. Previous studies focused on object detection and position estimation; our study, however, focused on detecting the direction of a person to enable the detection of a queue. Our results show that straight one-directional queues are detectable, but queues in different formations present problems.

In this study, we propose a method to apply the real-time object recognition using 360° spherical camera to web browser-based telecommunications. The main problems associated with the images captured by the spherical camera are as follows: i) the large geometric distortions of the objects near the poles in panoramic images projected by the equirectangular method and ii) the location of the object in the spherical image cannot be determined immediately. In addition, in conventional object detection methods, the browser-based, real-time object recognition method based on the use of a client-server system has not been established thus far. In this study, we addressed these problems by i) increasing the equirectangular panoramic images dataset, ii) training artificially distorted images of objects, and iii) adapting the object recognition model to Web Real-Time Communication. In our model, the range of object detection in 360° panoramic images expanded by approximately 25% compared with previous methods. Further, by transforming the results detected in a planar image into spherical coordinates, we have developed a system that can track a specific object in spherical images in real time. Our proposed system is not limited to any location as it can be utilized by users by simply inputting the image on the browser.

A spherical camera is a camera that can shoot omnidirectional landscapes in a celestial sphere. Images taken with a spherical camera can be converted using equirectangular projection, and 360° panoramic images can be viewed together in all directions. However, since distortion occurs in the upper and lower parts of the 360° panoramic image, it is difficult to recognize the object and grasp the position of the object even by looking at the image with the human eye. In this study, we propose a method to detect an object from a 360° panoramic image acquired from a spherical camera and estimate the position of the object from the detection result. By correcting the estimated position error, the accuracy was improved by about 2.5 times compared to before the correction.

Researchers must crawl geo-social data to analyze and visualize geo-social data. A conventional method to exhaustively crawl geosocial data is based on a grid. The crawler divides a specified area into a grid and uses the center coordinates of each cell to query databases using APIs. However, there is a difficult problem when using the grid-based method. It is that researchers cannot estimate the optimized grid size to exhaustively crawl geo-social data in advance because the optimized grid size depends on data density owing to geographical characteristics of an area. We focus on the fact that geo-social data are dense along roads. Thus, we propose a method based on road maps to exhaustively crawl geo-social data. We demonstrated that our method can crawl geo-social data by using almost the same number of queries compared to the crawler with an optimized grid size.

We propose a method that efficiently retrieves documents belonging to a target area from a georeferenced information database. A circular target area can be effectively searched by k-nearest neighbor searching, whereas a rectangular target is more appropriately searched by the bounding-box approach. However, queries of actual geographic information are often sourced from complex areas with different topographies and features. For efficient searching, queries on arbitrary and complex areas should be broken down into multiple simple queries, for example, a technique that divides the target area into grids. Our proposed method is a parameterless subquery construction method that efficiently searches georeferenced documents belonging to an arbitrary area specified by the user. Compared to the baseline method, the experimental results show that the number of subqueries could be reduced by up to 80%.

Multimedia Sensor Networks (MSNs) have gained increasing attention in recent years from both academic and industrial sectors. Unlike scalar sensor networks, the data collected from MSNs are enriched with multimedia data which can be used for defining and detecting complex and more application-meaningful events. However, to do so, there are several processing tasks need to be executed such as multimedia data decoding, translating semantic information from multimedia data, and integrating multimedia data from several sensors. Combining these tasks into one single generic framework so to process and detect complex events in MSNs is of great interest. However, developing such a framework is challenging due to the infrastructure of MSNs (which includes heterogeneous sensors) and types of multimedia data (which are diverse). Also, events in MSNs are needed to be detected in a near real-time manner. In this study, we propose an ontology-based framework to support complex event modeling and detecting in MSNs. Our framework helps users model MSNs infrastructure, complex events, and data collected from MSNs. It is also able of translating semantic formation, detecting, and reporting the events in a near real-time manner. Our framework is validated by means of prototyping and simulation. The results show that it can detect complex multimedia events in a high-work load scenario with average detection latency for less than 625 milliseconds.

Many photos shared on photo-sharing sites are annotated with tags and geo-tags. Some studies have demonstrated extraction of the geographical characterization which a tag represents as regions using those metadata. However, in some cases (e.g. coastline), a line is more suitable than a region as a geographical characterization of a tag. Therefore, we proposed a novel method to extract lines as a region as a geographical characterization. Results show that the distance of a coastline and many lines of our method is less than 500 m. Although, in this paper, only the coastline has been evaluated, this method is applicable to other tags as well.

A land cover map that represents the land surface of the earth is based primarily on analysis of remotely sensed images. However, the rate of concordance of existing land cover maps is not high. This lack of concordance results from a difference in classification methods and observation conditions of remotely sensed images. Also, conducting field surveys around the world is unrealistic. Therefore, we use ground level photographs from photo-sharing sites instead of field surveys. We propose a method to classify areas into land cover types using image features, geo-tags, titles and tags. Additionally, we create the land cover map using classified photographs. We evaluate the method using ground truth created manually. Results show that the accuracy of the proposed method is about 70 percent in New York.

Social media sites include many photos taken at various locations and times. In this paper, we propose a method to discover multiple hotspots, where many photos have been taken, using geo-tagging photos posted on social media sites. Additionally, we infer the range and shape of hotspots based on the deviation of the places where photos have been taken. We conducted experiments using image search results when we entered the search term "Rome" on Flickr. Our experimentally obtained results demonstrate that it is possible to discover multiple hotspots and to estimate the range and shape of each hotspot with high precision.