Person search aims at localizing and identifying a query person from a gallery of uncropped scene images. Different from person re-identification (re-ID), its performance also depends on the localization accuracy of a pedestrian detector. The state-of-the-art methods train the detector individually, and the detected bounding boxes may be suboptimal for the following re-ID task. To alleviate this issue, we propose a re-ID driven localization refinement framework for providing the refined detection boxes for person search. Specifically, we develop a differentiable ROI transform layer to effectively transform the bounding boxes from the original images. Thus, the box coordinates can be supervised by the re-ID training other than the original detection task. With this supervision, the detector can generate more reliable bounding boxes, and the downstream re-ID model can produce more discriminative embeddings based on the refined person localizations. Extensive experimental results on the widely used benchmarks demonstrate that our proposed method performs favorably against the stateof-the-art person search methods.
In the late 1970s and early 1980s, one of the post-Mao electoral reforms was semicompetitive elections, including those for local people's congresses. A better understanding of voters' subjective motivations in these elections is critical for explaining and predicting the significant effects of the elections on sociopolitical development in rapidly changing Chinese society. Using survey data collected in Beijing, China, in 1995, we reexamine arguments and findings about voters' subjective motivations reported by Shi (1999a). Contrary to Shi's arguments and findings, we find that people with stronger democratic orientation and a keener sense of internal efficacy are less likely to vote in these semicompetitive elections, while those who are identified with the regime and have affective attachments to the political authority are more likely to vote in the elections. In this article, we present the differences between our arguments and findings and Shi's. Then we draw some important political and theoretical implications from these differences.In the late 1970s and the early 1980s, the Chinese Communist Party (CCP) under Deng Xiaoping amended the electoral law for the election of people's congresses-Chinese "legislatures"-at various levels. Specifically, the new law introduced direct elections for local people's congresses. 1 According to this law, in theory voters could nominate candidates and have a choice among multiple candidates for each contested seat. Under the new law, these local elections have certainly become more competitive and transparent than those in the Mao era, but they are by no means fully competitive and democratic by any standard. These elections are still dominated and controlled by the CCP, which firmly upholds the one-party rule and allows only one official ideology (see e.g.
This article assumes that whether the current Chinese authoritarian government can maintain socio-political stability during the potentially turbulent transition to the post-Deng Xiaoping era depends, at least in part, upon the level of popular support for the political regime (or regime legitimacy). Based on data derived from a sample survey of Beijing residents, this study seeks to address two fundamental questions: “To what extent does the current Chinese communist regime enjoy public support?” and “What are the possible sources of popular support for the political regime in contemporary China?” The findings in this study suggest that (1) the current communist regime still enjoys a moderately high level of popular support, and (2) popular support for the regime is most likely to be found among those who are optimistic about the country's economic and political futures, who are most satisfied with their life, who give high evaluations of incumbent policies, who often follow public affairs, and who are older. Based on these findings, the article concludes that the current communist regime seems to have a good chance of remaining legitimate among a majority of the Chinese people, while it is still facing serious challenges from its policy performance in some major public policy areas.
In this paper, we propose a machine learning-based approach to detect malicious mobile malware in Android applications. This paper is able to capture instantaneous attacks that cannot be effectively detected in the past work. Based on the proposed approach, we implemented a malicious app detection tool, named Androidetect. First, we analyze the relationship between system functions, sensitive permissions, and sensitive application programming interfaces. The combination of system functions has been used to describe the application behaviors and construct eigenvectors. Subsequently, based on the eigenvectors, we compare the methodologies of naive Bayesian, J48 decision tree, and application functions decision algorithm regarding effective detection of malicious Android applications. Androidetect is then applied to test sample programs and real-world applications. The experimental results prove that Androidetect can better detect malicious applications of Android by using a combination of system functions compared with the previous work.INDEX TERMS Malicious applications of Android, machine learning, system function.
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.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.