Learning to recognize actions from only a handful of labeled videos is a challenging problem due to the scarcity of tediously collected activity labels. We approach this problem by learning a two-pathway temporal contrastive model using unlabeled videos at two different speeds leveraging the fact that changing video speed does not change an action. Specifically, we propose to maximize the similarity between encoded representations of the same video at two different speeds as well as minimize the similarity between different videos played at different speeds. This way we use the rich supervisory information in terms of 'time' that is present in otherwise unsupervised pool of videos. With this simple yet effective strategy of manipulating video playback rates, we considerably outperform video extensions of sophisticated state-of-the-art semi-supervised image recognition methods across multiple diverse benchmark datasets and network architectures. Interestingly, our proposed approach benefits from out-of-domain unlabeled videos showing generalization and robustness. We also perform rigorous ablations and analysis to validate our approach.
The location of disaster management facilities is a challenging and multifaceted problem. The road networks, population distribution along the road networks, and disaster risk maps are the major components for management of this problem. This article aims to decide on the location of such facilities for flood hazards in a region. The methodology is based on a multi‐objective framework. Objective functions include edge importance indices under fair weather and elevation parameters of the edges. Multiple scenarios are simulated for varying levels of hazard, and the outputs are analyzed. Analyses are carried out for the individual percentage loss of road links. A case study has been presented for the Bankura District in West Bengal, India. The inferences drawn from the results identify the critical links over the road networks of the region. The study also indicates locations in the region for relief facility setups to enable best‐serving capabilities and provide safe shelters, even in the most adverse flood conditions. The article depicts the vulnerability status of the road networks of the region. Further, it identifies the locations for relief facility provisioning that bring out the best road utilization and the best‐serving capabilities within the flood‐affected area under different flood levels.
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.