Search

Ex-ante online risk assessment for building emergency evacuation is essential to protect human life and property. Current risk assessment methods are limited by the tradeoff between accuracy and efficiency. In this paper, we propose an online method that overcomes this tradeoff based on multimedia data (e.g. videos data from surveillance cameras) and deep learning. The method consists of two parts. The first estimates the evacuee position as input for training the assessment model to then perform risk assessment in real scenarios. The second considers a social force model based on the evacuation simulation for the output of training model. We verify the proposed method in simulation and real scenarios. Model sensitivity analyses and large-scale tests demonstrate the usability and superiority of the proposed method. By the method, the computation time of risk assessment could be decreased from 10 minutes (by traditional simulation method) to 2.18 s. [ABSTRACT FROM AUTHOR]

Highlights from the article: Recently, however, people's mobile phone data, GPS trajectories data, location-based online social networking data, surveillance video data, satellite imagery and IC card data have become readily available and this information has increased explosively. B Memorial of Dr. Tao Li: b As the initial Lead Guest Editor of this Special Issue, Dr. Tao Li, a talented Professor of Computer Science at Florida International University, suffered a stroke during a doctoral defense on December 6, 2017.

We categorize this research in terms of its contribution to both graph theory and computer vision. From the theoretical perspective, this study can be considered as the first attempt to formulate the idea of mining maximal frequent subgraphs in the challenging domain of messy visual data, and as a conceptual extension to the unsupervised learning of graph matching. We define a soft attributed pattern (SAP) to represent the common subgraph pattern among a set of attributed relational graphs (ARGs), considering both their structure and attributes. Regarding the differences between ARGs with fuzzy attributes and conventional labeled graphs, we propose a new mining strategy that directly extracts the SAP with the maximal graph size without applying node enumeration. Given an initial graph template and a number of ARGs, we develop an unsupervised method to modify the graph template into the maximal-size SAP. From a practical perspective, this research develops a general platform for learning the category model (i.e., the SAP) from cluttered visual data (i.e., the ARGs) without labeling “what is where,” thereby opening the possibility for a series of applications in the era of big visual data. Experiments demonstrate the superior performance of the proposed method on RGB/RGB-D images and videos. [ABSTRACT FROM PUBLISHER]

Mining object-level knowledge, that is, building a comprehensive category model base, from a large set of cluttered scenes presents a considerable challenge to the field of artificial intelligence. How to initiate model learning with the least human supervision (i.e., manual labeling) and how to encode the structural knowledge are two elements of this challenge, as they largely determine the scalability and applicability of any solution. In this article, we propose a model-learning method that starts from a single-labeled object for each category, and mines further model knowledge from a number of informally captured, cluttered scenes. However, in these scenes, target objects are relatively small and have large variations in texture, scale, and rotation. Thus, to reduce the model bias normally associated with less supervised learning methods, we use the robust 3D shape in RGB-D images to guide our model learning, then apply the properly trained category models to both object detection and recognition in more conventional RGB images. In addition to model training for their own categories, the knowledge extracted from the RGB-D images can also be transferred to guide model learning for a new category, in which only RGB images without depth information in the new category are provided for training. Preliminary testing shows that the proposed method performs as well as fully supervised learning methods. [ABSTRACT FROM AUTHOR]

Online ride-hailing as an innovative travel mode becomes increasingly popular in cities around the world. To improve the efficiency of dispatching system of ride-hailing, many ride-hailing demand prediction models based on deep learning architecture have been proposed to reduce the gross travel distance between drivers and passengers. However, in most of these prediction models, only the error metrics are used for performance evaluation. There is scarce evidence on how much the demand prediction model reduces the unnecessary travel distance in reality. In this study, a multi-scenario-based method is proposed to evaluate technical potentials of the ride-hailing demand prediction model in ride-hailing dispatching system simulation. The ride-hailing dispatching is simulated in three scenarios: traditional dispatching system, prediction model-based dispatching system, and perfect prediction model-based dispatching system. One-month data of Didi Express service provided by Didi Chuxing GAIA Initiative in Chengdu is employed to support the simulation. Two terms, empty distance and relative performance, are introduced as the criteria of prediction model performance measurement. Simulation results reveal that the total empty distance reduced 1,164 km per day by using the prediction model compared with the traditional dispatching system. The relative performance is only 58.9% compared with the perfect prediction model-based dispatching system. [ABSTRACT FROM AUTHOR]

The on-road braking emission has been proved by former studies to account for a considerable part of on-road transportation. To improve cleaner air in urban area, the spatio-temporal analysis on emission performance of on-road braking is necessary as a guideline for decision-making. In this paper, we propose a framework for analysis on the particle matter emission of on-road vehicle braking events on an urban scale. We used massive vehicle trajectories in Tokyo area with short time interval as the database for analysis. From the result, we found that within the in the study area, the average driving distance during braking takes up about 20.60% of total driving distance. The average quantity of PM10 emission from braking for each driving trajectory is 14.09 mg and the one from exhaust emission is 35.36 mg. The emission from braking can averagely occupy 39.85% of exhaust emission. What's more, in our finding, the braking emission from heavy duty vehicle is 2.33 times of light duty vehicle. From the spatial distribution of PM10 braking emission, we found that braking emission usually happened in the city center and popular crowded areas due to the large traffic volume, as well as the main trunk roads of capital expressway or highway. We also found a different spatial pattern between the light duty vehicle and heavy-duty vehicle. In temporal change, we found that rapid peaks during the rush hour on weekday and contrastive stabilization on weekend. We believe our finding can provide a clearer pattern and understanding on the emission behavior of on-road vehicle braking. Image 1 • Proposing a framework to for analysis on the particle matter emission of on-road vehicle braking events on an urban scale. • Proposing a method to detect and mine emission from braking behaviors from GPS trajectory data of vehicle travelling. • Over twenty million historical GPS points of individual vehicle in Tokyo area are employed for analysis. • We found that emission from braking can occupy 39.85% of exhaust emission and HDV emits 2.33 times of LDV. [ABSTRACT FROM AUTHOR]