Describing user activity plays an essential role in ambient intelligence. In this work, we review different methods for human activity recognition, classified as data-driven and knowledge-based techniques. We focus on context ontologies whose ultimate goal is the tracking of human behavior. After studying upper and domain ontologies, both useful for human activity representation and inference, we establish an evaluation criterion to assess the suitability of the different candidate ontologies for this purpose. As a result, any missing features, which are relevant for modeling daily human behaviors, are identified as future challenges.
Availability of domain-specific datasets is an essential problem in object detection. Datasets of inshore and offshore maritime vessels are no exception, with a limited number of studies addressing maritime vessel detection on such datasets. For that reason, we collected a dataset consisting of images of maritime vessels taking into account different factors: background variation, atmospheric conditions, illumination, visible proportion, occlusion and scale variation. Vessel instances (including nine types of vessels), seamarks and miscellaneous floaters were precisely annotated: we employed a first round of labelling and we subsequently used the CSRT tracker to trace inconsistencies and relabel inadequate label instances. Moreover, we evaluated the out-of-the-box performance of four prevalent object detection algorithms (Faster R-CNN, R-FCN, SSD and EfficientDet). The algorithms were previously trained on the Microsoft COCO dataset. We compared their accuracy based on feature extractor and object size. Our experiments showed that Faster R-CNN with Inception-Resnet v2 outperforms the other algorithms, except in the large object category where EfficientDet surpasses the latter.
This paper presents prediction-based dynamic resource allocation algorithms to scale video transcoding service on a given Infrastructure as a Service cloud. The proposed algorithms provide mechanisms for allocation and deallocation of virtual machines (VMs) to a cluster of video transcoding servers in a horizontal fashion. We use a two-step load prediction method, which allows proactive resource allocation with high prediction accuracy under real-time constraints. For cost-efficiency, our work supports transcoding of multiple on-demand video streams concurrently on a single VM, resulting in a reduced number of required VMs. We use video segmentation at group of pictures level, which splits video streams into smaller segments that can be transcoded independently of one another. The approach is demonstrated in a discrete-event simulation and an experimental evaluation involving two different load patterns.