Object tracking is an important computer vision task that has drawn increasing attention because of its academic and commercial potential. Although several trackers based on convolutional neural network (CNN) have shown strong performance in single object tracking and multiple object tracking, it remains challenging to perform real-time object tracking on videos with frame rate over 30 frame per second. For real-world applications, one of the necessary requirements is the trackers need to estimate the bounding boxes with a rate higher or equal to the video frame rate to avoid latency. In this study, we propose a hybrid tracker combining the high accuracy of CNN-based trackers and the superior processing speed of particle filtering algorithm to achieve real-time tracking. The experimental results indicate our tracker achieves a state-of-the-art performance on the benchmark multi-object tracking datasets with competitive accuracy.
In the context of smart city, there is the need of cheap approaches to get environmental and contextual data. IoT (Internet of Things) devices and sensors may be located in specific points of interest, while their costs is high for the installation, usage/licensing and maintenance. The satellite data can be a viable solution to have more data at the cheaper costs, some of them are open data, while in other cases the satellite data have a cost. In this paper, the exploitation of satellite data in the context of Smart City is analyzed and some examples are provided. The satellite data of the European Union’s Earth observation program Copernicus can be used to calibrate the values of large sensors network data, and for new applications in which similar data cannot be recovered from the field. A demonstrative Dashboard has been provided to allow the users to perform a comparison of data coming from satellite with respect to those obtained from IOT Devices. This research has been developed in the context of PC4City project (Civil Protection for the City) of the Fondazione Cassa di Risparmio di Firenze.
Process visualizations of data from manufacturing execution systems (MESs) provide the ability to generate valuable insights for improved decision-making. Industry 4.0 is awakening a digital transformation where advanced analytics and visualizations are critical. Exploiting MESs with data-driven strategies can have a major impact on business outcomes. The advantages of employing process visualizations are demonstrated through an application to real-world data. Visualizations, such as dashboards, enable the user to examine the performance of a production line at a high level. Furthermore, the addition of interactivity facilitates the user to customize the data they want to observe. Evidence of process variability between shifts and days of the week can be investigated with the goal of optimizing production.
Making lane-changing decisions for autonomous vehicles has been widely regarded as a core task for autonomous driving. In this paper, we proposed an imitation learning framework to navigate a target vehicle through congested highway scenarios. In this framework, to efficiently capture the dynamic traffic environments, a representative feature named Gaussian Velocity field which is invariant to the number of surrounding obstacles is introduced. Then we train a GVF-CNN model to determine the actions of the target vehicle among a set of five categories which helps autonomous vehicles make decisions. Simulations are conducted in top-down bird-eye view and demonstration data are collected from two ways: (a) highD dataset, a real-world collected highway dataset for pre-training the model, and (b) simulation data for fine-tuning the model. By adopting the well-trained model to control the target vehicle in the simulator to conduct the simulation, the experiment results demonstrate the effectiveness of our method. Finally, detailed analysis of crash reasons as well as the potential improvement methods are discussed.
Unmanned aerial vehicle (UAV)-assisted data collection in Internet of Things (IoT) networks is an emerging and promising paradigm in the next generation communication systems. This paper considers a scenario where a UAV collects data from multiple IoT sensor equipments (SEs) one by one and finally flies to the data center for data processing. To reduce the energy consumption of the SEs, backscatter communication is adopted. The paper aims to minimize the average age of information (AoI) of all SEs by jointly optimizing the UAV’s data collection time from different SEs, the transmission power of the UAV, and the trajectory of the UAV. Since the optimization problem includes both discrete variables (the collection sequence of SEs) and continuous variables (the transmission power and collection time), the problem is difficult to solve by conventional methods. By exploiting the special structure characteristics of the problem, it can be divided into two sub-problems without loss of optimality. The first one is the data collection time minimization problem that contains only continuous variables, and its optimal solution is derived based on the KKT optimality conditions. While the second one contains only the discrete variables, and a backtracking scheme based UAV trajectory design method is proposed. Numerical results verify that the proposed method can always achieve significantly better performance in terms of the average AoI than three benchmark algorithms.
