•A Comprehensive study on the BoVW pipeline for action recognition task.•An evaluation and a generic analysis on 13 encoding methods.•Intra-normalization for supervector based encoding methods.•An ...evaluation on three fusion methods.•Several good practices of the BoVW pipeline for action recognition task.
Video based action recognition is one of the important and challenging problems in computer vision research. Bag of visual words model (BoVW) with local features has been very popular for a long time and obtained the state-of-the-art performance on several realistic datasets, such as the HMDB51, UCF50, and UCF101. BoVW is a general pipeline to construct a global representation from local features, which is mainly composed of five steps; (i) feature extraction, (ii) feature pre-processing, (iii) codebook generation, (iv) feature encoding, and (v) pooling and normalization. Although many efforts have been made in each step independently in different scenarios, their effects on action recognition are still unknown. Meanwhile, video data exhibits different views of visual patterns , such as static appearance and motion dynamics. Multiple descriptors are usually extracted to represent these different views. Fusing these descriptors is crucial for boosting the final performance of an action recognition system. This paper aims to provide a comprehensive study of all steps in BoVW and different fusion methods, and uncover some good practices to produce a state-of-the-art action recognition system. Specifically, we explore two kinds of local features, ten kinds of encoding methods, eight kinds of pooling and normalization strategies, and three kinds of fusion methods. We conclude that every step is crucial for contributing to the final recognition rate and improper choice in one of the steps may counteract the performance improvement of other steps. Furthermore, based on our comprehensive study, we propose a simple yet effective representation, called hybrid supervector, by exploring the complementarity of different BoVW frameworks with improved dense trajectories. Using this representation, we obtain impressive results on the three challenging datasets; HMDB51 (61.9%), UCF50 (92.3%), and UCF101 (87.9%).
Full text
Available for:
GEOZS, IJS, IMTLJ, KISLJ, NUK, OILJ, SAZU, SBCE, UL, UPUK
Temporal Segment Networks for Action Recognition in Videos Wang, Limin; Xiong, Yuanjun; Wang, Zhe ...
IEEE transactions on pattern analysis and machine intelligence,
2019-Nov.-1, 2019-Nov, 2019-11-1, 20191101, Volume:
41, Issue:
11
Journal Article
Peer reviewed
Open access
We present a general and flexible video-level framework for learning action models in videos. This method, called temporal segment network (TSN), aims to model long-range temporal structure with a ...new segment-based sampling and aggregation scheme. This unique design enables the TSN framework to efficiently learn action models by using the whole video. The learned models could be easily deployed for action recognition in both trimmed and untrimmed videos with simple average pooling and multi-scale temporal window integration, respectively. We also study a series of good practices for the implementation of the TSN framework given limited training samples. Our approach obtains the state-the-of-art performance on five challenging action recognition benchmarks: HMDB51 (71.0 percent), UCF101 (94.9 percent), THUMOS14 (80.1 percent), ActivityNet v1.2 (89.6 percent), and Kinetics400 (75.7 percent). In addition, using the proposed RGB difference as a simple motion representation, our method can still achieve competitive accuracy on UCF101 (91.0 percent) while running at 340 FPS. Furthermore, based on the proposed TSN framework, we won the video classification track at the ActivityNet challenge 2016 among 24 teams.
Application of transition metal phosphides (TMPs) for electrochemical energy conversion and storage has great potential to alleviate the energy crisis. Although there are many methods to get TMPs, it ...is still immensely challenging to fabricate hierarchical porous TMPs with superior electrochemical performances by a simple, green, and secure approach. Herein, we report a facile method to synthesize the CoP/C nanoboxes by pyrolysis of phytic acid (PA) cross-linked Co complexes that are acquired from reaction of PA and ZIF-67. The PA can not only slowly etch ZIF-67 and gain a hollow structure but also act as a source of phosphorus to prepare CoP/C nanoboxes. The CoP/C nanoboxes deliver an ultrahigh specific capacity (868 mA h g–1 at 100 mA g–1) and excellent cycle stability (523 mA h g–1 after 1000 cycles at 500 mA h g–1) when used as anode materials for lithium-ion batteries. Moreover, when used as an electrocatalyst for hydrogen evolution reaction, the CoP/C nanoboxes exhibit ultralow overpotential, small Tafel slope, and excellent durability in acidic media. The method to produce CoP/C nanoboxes is easy and environmentally friendly and can be readily extended to design other TMPs/C nanocomposites.
Full text
Available for:
IJS, KILJ, NUK, PNG, UL, UM
Currently, research on the Chinese legal system is flourishing, and more and more issues are worthy of exploration. The Chinese legal system is unique among the five major legal systems in the world. ...It is the only ancient secular legal system, and the only ancient secular legal system to have been disseminated by non-coercive means. The openness of the Chinese legal system is a powerful proof that traditional Chinese law plays a role in importing and exporting legal policies and measures from and to other states. Indeed, these issues are important in research on the Chinese legal system, and a correct understanding of them can help to comprehensively understand traditional Chinese legal culture. It is essential to deepen the research on the Chinese legal system to contribute to its revival and to the inheritance of traditional Chinese legal culture. In addition to comprehensive discussions, it is also necessary to fill gaps and advance the research on the Chinese legal system.
Full text
Available for:
CEKLJ, NUK, ODKLJ, PRFLJ, UM
Cancer tissues consist of cancer cells, surrounding stromal cells and the extracellular matrix. Cancer‐associated fibroblasts (CAF) are one of the key components of stromal cells. CAF have a great ...impact on the behavior of cancer cells, including proliferation, invasion, metastasis and chemoresistance in many ways. However, the underlying mechanism had not been fully elucidated. In this study, we investigated the role of CAF in cisplatin resistance of lung cancer cells. By using conditioned medium from CAF (CAF‐CM), we found that CAF decreased the sensitivity of lung cancer cells to cisplatin. RNA sequencing results showed that CAF expressed a higher level of Annexin A3 (ANXA3) than normal fibroblasts (NF), and CAF‐CM incubation increased the ANXA3 level in lung cancer cells. Overexpression of ANXA3 in lung cancer cells increased cisplatin resistance and activated c‐jun N‐terminal kinase (JNK), whereas knockdown of ANXA3 increased cisplatin sensitivity. Further study showed that CAF‐CM enhanced cisplatin resistance by inhibiting cisplatin‐induced apoptosis, determined by repression of caspase‐3 and caspase‐8, through activation of the ANXA3/JNK pathway. Conversely, suppression of JNK activation by specific inhibitor retarded the effect of CAF‐CM and ANXA3 on cisplatin sensitivity. Taken together, our study demonstrated that CAF potentiated chemoresistance of lung cancer cells through a novel ANXA3/JNK pathway both in vitro and in vivo, suggesting ANXA3 could be a potential therapeutic target for the treatment of chemoresistant cancer.
Cancer‐associated fibroblasts decreased the sensitivity of lung cancer cells to cisplatin. CAF expressed higher level of ANXA3 than normal fibroblasts (NF), and CAF‐CM incubation increased the ANXA3 level in lung cancer cells. CAF potentiated chemoresistance of lung cancer cells through a novel ANXA3/JNK pathway both in vitro and in vivo.
Full text
Available for:
BFBNIB, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK
Visual features are of vital importance for human action understanding in videos. This paper presents a new video representation, called trajectory-pooled deep-convolutional descriptor (TDD), which ...shares the merits of both hand-crafted features 31 and deep-learned features 24. Specifically, we utilize deep architectures to learn discriminative convolutional feature maps, and conduct trajectory-constrained pooling to aggregate these convolutional features into effective descriptors. To enhance the robustness of TDDs, we design two normalization methods to transform convolutional feature maps, namely spatiotemporal normalization and channel normalization. The advantages of our features come from (i) TDDs are automatically learned and contain high discriminative capacity compared with those hand-crafted features; (ii) TDDs take account of the intrinsic characteristics of temporal dimension and introduce the strategies of trajectory-constrained sampling and pooling for aggregating deep-learned features. We conduct experiments on two challenging datasets: HMD-B51 and UCF101. Experimental results show that TDDs outperform previous hand-crafted features 31 and deep-learned features 24. Our method also achieves superior performance to the state of the art on these datasets.
Electron-hole separation is a critical step to achieving efficient photocatalysis, towards which use of co-catalysts has become a widely used strategy. Despite the tremendous efforts and demonstrated ...functions of noble metal co-catalysts, seeking noble metal-free co-catalysts will always be the goal when designing cost- effective, high-performance hybrid photocatalysts. In this work, we demonstrate that MoS~ nanosheets with 1T phase （i.e., octahedral phase） can function as a co-catalyst with multiple merits： （1） Noble-metal-free; （2） high mobility for charge transport; （3） high density of active sites for H2 evolution on basal planes; （4） good performance stability; （5） high light transparency. As demonstrated in both photocatalytic hydrogen production and Rhodamine B degradation, the developed hybrid structure with TiO2 exhibits excellent performance, in sharp contrast to bare TiO2 and the hybrid counterpart with 2H-MoS2.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
Hybridizing nanostructured metal oxides with multiwalled carbon nanotubes (MWCNTs) is highly desirable for the improvement of electrochemical performance of lithium-ion batteries. Here, a facile and ...scalable strategy to fabricate hierarchical porous MWCNTs/Co3O4 nanocomposites has been reported, with the help of a morphology-maintained annealing treatment of carbon nanotubes inserted metal organic frameworks (MOFs). The designed MWCNTs/Co3O4 integrates the high theoretical capacity of Co3O4 and excellent conductivity as well as strong mechanical/chemical stability of MWCNTs. When tested as anode materials for lithium-ion batteries, the nanocomposite displays a high reversible capacity of 813 mAh g–1 at a current density of 100 mA g–1 after 100 charge–discharge cycles. Even at 1000 mA g–1, a stable capacity as high as 514 mAh g–1 could be maintained. The improved reversible capacity, excellent cycling stability, and good rate capability of MWCNTs/Co3O4 can be attributed to the hierarchical porous structure and the synergistic effect between Co3O4 and MWCNTs. Furthermore, owing to this versatile strategy, binary metal oxides MWCNTs/ZnCo2O4 could also be synthesized as promising anode materials for advanced lithium-ion batteries.
Full text
Available for:
IJS, KILJ, NUK, PNG, UL, UM
Rice cultivation usually involves high water and fertilizer application rates leading to the nonpoint pollution of surface waters with phosphorus (P) and nitrogen (N). Here, a 10-year field ...experiment was conducted to investigate N and P losses and their impact factors under different irrigation and fertilization regimes. Results indicated that T2 (Chemical fertilizer of 240 kg N ha.sup.-1, 52 kg P ha.sup.-1, and 198 kg K ha.sup.-1 combined with shallow intermittent irrigation) decreased N loss by 48.9% compared with T1 (Chemical fertilizer of 273 kg N ha.sup.-1, 59 kg P ha.sup.-1, and 112 kg K ha.sup.-1 combined with traditional flooding irrigation). The loss ratio (total N loss loading/amount of applied N) of N was 9.24-15.90%, whereas that of P was 1.13-1.31% in all treatments. Nitrate N (NO.sub.3 .sup.- -N) loss was the major proportion accounting for 88.30-90.65% of dissolved inorganic N loss through surface runoff. Moreover, the N runoff loss was mainly due to high fertilizer input, soil NO.sub.3 .sup.- -N, and ammonium N (NH.sub.4 .sup.+ -N) contents. In addition, the N loss was accelerated by Bacteroidetes, Proteobacteria, Planotomycetes, Nitrospirae, Firmicutes bacteria and Ascomycota fungi, but decreased by Chytridiomycota fungi whose contribution to the N transformation process. Furthermore, T2 increased agronomic N use efficiency (AEN) and rice yield by 32.81% and 7.36%, respectively, in comparison with T1. These findings demonstrated that T2 might be an effective approach to ameliorate soil chemical properties, regulate microbial community structure, increase AEN and consequently reduce N losses as well as maintaining rice yields in the present study.
Full text
Available for:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
IMPORTANCE: Previous studies have shown increasing prevalence of diabetes in China, which now has the world’s largest diabetes epidemic. OBJECTIVES: To estimate the recent prevalence and to ...investigate the ethnic variation of diabetes and prediabetes in the Chinese adult population. DESIGN, SETTING, AND PARTICIPANTS: A nationally representative cross-sectional survey in 2013 in mainland China, which consisted of 170 287 participants. EXPOSURES: Fasting plasma glucose and hemoglobin A1c levels were measured for all participants. A 2-hour oral glucose tolerance test was conducted for all participants without diagnosed diabetes. MAIN OUTCOMES AND MEASURES: Primary outcomes were total diabetes and prediabetes defined according to the 2010 American Diabetes Association criteria. Awareness and treatment were also evaluated. Hemoglobin A1c concentration of less than 7.0% among treated diabetes patients was considered adequate glycemic control. Minority ethnic groups in China with at least 1000 participants (Tibetan, Zhuang, Manchu, Uyghur, and Muslim) were compared with Han participants. RESULTS: Among the Chinese adult population, the estimated standardized prevalence of total diagnosed and undiagnosed diabetes was 10.9% (95% CI, 10.4%-11.5%); that of diagnosed diabetes, 4.0% (95% CI, 3.6%-4.3%); and that of prediabetes, 35.7% (95% CI, 34.1%-37.4%). Among persons with diabetes, 36.5% (95% CI, 34.3%-38.6%) were aware of their diagnosis and 32.2% (95% CI, 30.1%-34.2%) were treated; 49.2% (95% CI, 46.9%-51.5%) of patients treated had adequate glycemic control. Tibetan and Muslim Chinese had significantly lower crude prevalence of diabetes than Han participants (14.7% 95% CI, 14.6%-14.9% for Han, 4.3% 95% CI, 3.5%-5.0% for Tibetan, and 10.6% 95% CI, 9.3%-11.9% for Muslim; P < .001 for Tibetan and Muslim compared with Han). In the multivariable logistic models, the adjusted odds ratios compared with Han participants were 0.42 (95% CI, 0.35-0.50) for diabetes and 0.77 (95% CI, 0.71-0.84) for prediabetes for Tibetan Chinese and 0.73 (95% CI, 0.63-0.85) for diabetes and 0.78 (95% CI, 0.71-0.86) for prediabetes in Muslim Chinese. CONCLUSIONS AND RELEVANCE: Among adults in China, the estimated overall prevalence of diabetes was 10.9%, and that for prediabetes was 35.7%. Differences from previous estimates for 2010 may be due to an alternate method of measuring hemoglobin A1c.
PDF
