Landslide is one of the most dangerous disasters, especially for countries with large mountainous terrain. It causes a great damage to lives, infrastructure and environments, such as traffic congestion and high accidents. Therefore, automated landslide detection is an important task for warning and reducing its consequences such as blocked traffic or traffic accidents. For instance, people approaching the disaster area can adjust their routes to avoid blocked roads, or dangerous traffic signs can be positioned in time to warn the traffic participants to avoid the interrupted road ahead. This paper proposes a method to detect blocked roads caused by landslide by utilizing images captured from Unmanned Aerial Vehicles (UAV). The proposed method comprises of three components: road segmentation, blocked road candidate extraction, and blocked road classification, which is leveraged by a multi-stage convolutional neural network model. Our experiments demonstrate that the proposed method can surpass over several state-of-the art methods on our self-collected dataset of 400 images captured with an UAV.
In the era of information explosion, a program is necessary to be scalable. Therefore, scalability analysis becomes very important in software verification and validation. However, current approaches to empirical scalability analysis remain limitations related to the number of supported models and performance. In this paper, we propose a runtime approach for estimating the program resource usage with two aims: evaluating the program scalability and revealing potential errors. In this approach, the resource usage of a program is first observed when it is executed on inputs with different scales, the observed results are then fitted on a model of the usage according to the program's input. Comparing to other approaches, ours supports diverse models to illustrate the resource usage, i.e., linear-log, power-law, polynomial, etc. We currently focus on the computation cost and stack frames usage as two representatives of resource usage, but the approach can be extended to other kinds of resource. The experimental result shows that our approach achieves more precise estimation and better performance than other state-of-the-art approaches.
Abstract: In speech synthesis and recognition, the segmentation is an important step. The result of further steps depend completely on this process. There are several effective segmentation method in the literature,
but for Vietnamese speech, researchers usually base on their experience to set the length while using sliding window. It causes an inefficient segmentation; and they need to try with the other value (length of voice). In this paper, we propose a method supporting in segmentation for Vietnamese speech and automatically determine the suitable length of voices and silent pause. We firstly estimate, by experimenting, the min and average length of a voice and a silent pause for Vietnamese speech in three main type speaking (slow, normal and fast). Then, based on these values, we start to segment the voice and pause by sliding window with proposed algorithm. Experiment results show that the proposed method can be used to effectively segment the Vietnamese speech.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.