17 research outputs found
Surface Based Object Detection in RGBD Images
No full textInternational audienceViewpoint variation is one of the main challenges for object-detection frameworks. In this work we describe strategies to improve object detection pipelines by introducing viewpoint based mixture components. We learn accurate mixtures of object detectors for RGB-Depth (RGBD) data using the latent SVM framework. Our contributions are threefold. First, we use surface-based object representations (3D mesh models) from available 3D object model repositories to learn strongly supervised viewpoint classifiers. These are used to guide the first stages of model learning, and help avoid inaccurate local minima of latent SVM training. Second, we develop a geometric dataset augmentation scheme that uses scene geometry to 'take another look' at the training data, simulating the effect of camera viewpoint changes. Third, to better exploit depth information, we develop a novel depth-based dense feature extraction method that provides a robust statistical description of scene geometry. We evaluate our learned detectors on the NYU dataset, and demonstrate that each of our advances results in systematic performance improvements over the traditional HOG-based detection pipeline
Offline Deformable Face Tracking in Arbitrary Videos
No full textGeneric face detection and facial landmark localization in static imagery are among the most mature and well-studied problems in machine learning and computer vision. Currently, the top performing face detectors achieve a true positive rate of around 75-80% whilst maintaining low false positive rates. Furthermore, the top performing facial landmark localization algorithms obtain low point-to-point errors for more than 70% of commonly benchmarked images captured under unconstrained conditions. The task of facial landmark tracking in videos, however, has attracted much less attention. Generally, a tracking-by-detection framework is applied, where face detection and landmark localization are employed in every frame in order to avoid drifting. Thus, this solution is equivalent to landmark detection in static imagery. Empirically, a straightforward application of such a framework cannot achieve higher performance, on average, than the one reported for static imagery. In this paper, we show for the first time, to the best of our knowledge, that the results of generic face detection and landmark localization can be used to recursively train powerful and accurate person-specific face detectors and landmark localization methods for offline deformable tracking. The proposed pipeline can track landmarks in very challenging long-term sequences captured under arbitrary conditions. The pipeline was used as a semi-automatic tool to annotate the majority of the videos of the 300-VW Challenge
Author Correction: Early treatment of COVID-19 with anakinra guided by soluble urokinase plasminogen receptor plasma levels: a double-blind, randomized controlled phase 3 trial (Nature Medicine, (2021), 27, 10, (1752-1760), 10.1038/s41591-021-01499-z)
No full textIn the version of this Article initially published, there was an error in the author affiliations. Specifically, affiliation 27, corresponding to author Carlo Selmi, has been corrected from “Humanitas Research Hospital, Milan, Italy” to read: “Department of Biomedical Sciences, Humanitas University, Milan, Italy & IRCCS Humanitas Research Hospital, Milan, Italy.” The change has been made to the online version of the Article
The first Facial Landmark Tracking in-the-Wild Challenge: benchmark and results
No full textDetection and tracking of faces in image sequences is among the most well studied problems in the intersection of statistical machine learning and computer vision. Often, tracking and detection methodologies use a rigid representation to describe the facial region 1, hence they can neither capture nor exploit the non-rigid facial deformations, which are crucial for countless of applications (e.g., facial expression analysis, facial motion capture, high-performance face recognition etc.). Usually, the non-rigid deformations are captured by locating and tracking the position of a set of fiducial facial landmarks (e.g., eyes, nose, mouth etc.). Recently, we witnessed a burst of research in automatic facial landmark localisation in static imagery. This is partly attributed to the availability of large amount of annotated data, many of which have been provided by the first facial landmark localisation challenge (also known as 300-W challenge). Even though now well established benchmarks exist for facial landmark localisation in static imagery, to the best of our knowledge, there is no established benchmark for assessing the performance of facial landmark tracking methodologies, containing an adequate number of annotated face videos. In conjunction with ICCV’2015 we run the first competition/challenge on facial landmark tracking in long-term videos. In this paper, we present the first benchmark for long-term facial landmark tracking, containing currently over 110 annotated videos, and we summarise the results of the competition
Author Correction: Early treatment of COVID-19 with anakinra guided by soluble urokinase plasminogen receptor plasma levels: a double-blind, randomized controlled phase 3 trial (Nature Medicine, (2021), 27, 10, (1752-1760), 10.1038/s41591-021-01499-z)
No full textIn the version of this Article initially published, there was an error in the author affiliations. Specifically, affiliation 27, corresponding to author Carlo Selmi, has been corrected from “Humanitas Research Hospital, Milan, Italy” to read: “Department of Biomedical Sciences, Humanitas University, Milan, Italy & IRCCS Humanitas Research Hospital, Milan, Italy.” The change has been made to the online version of the Article. © The Author(s) 2021
HIV cascade of care in Greece: Useful insights from additional stages
No full textBackground Aiming to eliminate HIV infection, UNAIDS has set a global “90-90-90” target by 2020. We sought to construct a 6-stages HIV Cascade of Care (CoC) in Greece, overall and by risk group, to assess risk-group and stage-specific progress in achieving the UNAIDS target. Patients and methods Combining data from the HIV/AIDS surveillance system and a population-based HIV cohort study, the CoC included: i) number of people living with HIV (PLHIV) by end of 2013; ii) proportion of PLHIV ever diagnosed; iii) proportion of diagnosed linked-to-care iv) proportion of linked-to-care ever initiating antiretroviral therapy (ART); v) proportion of treated who retained-in-care vi) proportion of those retained-in-care who were virally suppressed (200 copies/mL) at their last visit (01/07/2012-31/12/2013). Results In 2013, 14147 PLHIV were in Greece. Overall, proportions of each stage in the cascade were: 78.4% diagnosed; 86% linked-to-care; 78.5% initiated ART; 86.4% retained-in-care, and 87.1% virally suppressed. Totally, 42.6% of all PLHIV were virally suppressed. The percentage diagnosed was lower among heterosexual men and women (heterosexuals) than in MSM (men who have sex with men) or PWID (people who inject drugs). Most MSM were linked to care (97.2% of diagnosed) while a substantial proportion of PWID were not (80.8% of diagnosed). Once treated, PWID remained in care in similar proportions to MSM. Unlike PWID, a high proportion of the retained in care MSM and heterosexuals achieved viral suppression. Conclusions At the end of 2013, we identified gaps in the HIV CoC in Greece, which differed across risk groups. Targeted interventions are critical in optimizing early diagnosis and timely linkage. A 6-stage CoC, stratified by risk group, can inform strategic public health planning in improving HIV treatment outcomes. © 2018 Vourli et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Low pre-ART CD4 count is associated with increased risk of clinical progression or death even after reaching 500 CD4 cells/μL on ART
No full textIntroduction Clinical disadvantages of initiating ART at low CD4 counts have been clearly demonstrated but whether any excess risk remains even after reaching relatively high/safe CD4 levels remains unclear. We explore whether individuals starting ART with <500 CD4 cells/μL who increased their CD4 count above this level, have, from this point onwards, similar risk of clinical progression to serious AIDS/non-AIDS events or death with individuals starting ART with ≥500 CD4 cells/μL. Methods Data were derived from a multicenter cohort (AMACS). Adults, starting PI, NNRTI or INSTI based ART, in or after 2000 were eligible, provided they started ART with ≥500 (“High CD4”) or started with CD4 <500 cells/μL but surpassed this threshold while on ART (“Low CD4”). Baseline was the date of ART initiation (“High CD4”) or of first reaching 500 CD4 cells/μL (“Low CD4”). Survival analysis, allowing for competing risks, was used to explore the risk of progression to study’s endpoints. Results The study included 694 persons in the “High CD4” and 3,306 in the “Low CD4” group. Median (IQR) follow-up was 66 (36, 106) months. In total, 257 events (40 AIDS related, 217 SNAEs) were observed. Rates of progression did not differ significantly between the two groups but the subgroup of those initiating ART with <200 CD4 cells/μL had significantly higher risk of progression after baseline, compared to those in the “High CD4” group. Conclusions Individuals starting ART with <200 cells/μL remain on increased risk even after reaching 500 CD4 cells/μL. These patients should be closely followed. Copyright: © 2023 Pantazis et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
The HIV patient profile in 2013 and 2003: Results from the Greek AMACS cohort
No full textCombined Antiretroviral therapy (cART) has improved life-expectancy of people living with HIV (PLHIV) but as they age, prevalence of chronic non-AIDS related comorbidities may increase. We study the evolution of HIV-disease markers and comorbidities’ prevalence in PLHIV in Greece. Two cross-sectional analyses (2003 and 2013) on data from the AMACS cohort were performed. Comparisons were based on population average models and were repeated for subjects under follow-up at both 2003 and 2013. 2,403 PLHIV were identified in 2003 and 4,910 in 2013 (1,730 contributing for both cross-sections). Individuals in 2013 were on average older, diagnosed/treated for HIV for longer, more likely to be on cART, viro-logically suppressed, and with higher CD4 counts. Chronic kidney disease, dyslipidemia and hypertension prevalence increased over time. There was an increase in prescription of lipid-lowering treatment (3.5% in 2003 vs. 7.7% 2013, p<0.001). Among 220 and 879 individuals eligible for Framingham 10-year Event Risk calculation, the proportion of patients in the high-risk group (>20%) increased from 18.2% to 22.2% (p = 0.002). Increase in the prevalence of comorbidities was more pronounced in the subset of patients who were followed in both 2003 and 2013. The increased availability and uptake of cART led to significant improvements in the immuno-virological status of PLHIV in Greece, but they aged alongside an increase in prevalence of non-AIDS related comorbidities. These results highlight the need for appropriate monitoring, optimal cART selection and long-term management and prevention strategies for such comorbidities. © 2018 Pantazis et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Ceftazidime/avibactam in the era of carbapenemase-producing Klebsiella pneumoniae: Experience from a national registry study
No full textBackground: Infections caused by KPC-producing Klebsiella pneumoniae (Kp) are associated with high mortality. Therefore, new treatment options are urgently required. Objectives: To assess the outcomes and predictors of mortality in patients with KPC-or OXA-48-Kp infections treated with ceftazidime/avibactam with an emphasis on KPC-Kp bloodstream infections (BSIs). Methods: A multicentre prospective observational study was conducted between January 2018 and March 2019. Patients with KPC-or OXA-48-Kp infections treated with ceftazidime/avibactam were included in the analysis. The subgroup of patients with KPC-Kp BSIs treated with ceftazidime/avibactam was matched by propensity score with a cohort of patients whose KPC-Kp BSIs had been treated with agents other than ceftazidime/avibactam with in vitro activity. Results: One hundred and forty-seven patients were identified; 140 were infected with KPC producers and 7 with OXA-48 producers. For targeted therapy, 68 (46.3%) patients received monotherapy with ceftazidime/avibactam and 79 (53.7%) patients received ceftazidime/avibactam in combination with at least another active agent. The 14 and 28 day mortality rates were 9% and 20%, respectively. The 28 day mortality among the 71 patients with KPC-Kp BSIs treated with ceftazidime/avibactam was significantly lower than that observed in the 71 matched patients, whose KPC-Kp BSIs had been treated with agents other than ceftazidime/avibactam (18.3% versus 40.8%; P = 0.005). In the Cox proportional hazards model, ultimately fatal disease, rapidly fatal disease and Charlson comorbidity index ≥2 were independent predictors of death, whereas treatment with ceftazidime/avibactam-containing regimens was the only independent predictor of survival. Conclusions: Ceftazidime/avibactam appears to be an effective treatment against serious infections caused by KPC-Kp. © 2021 The Author(s)
The potential of using dynamic information flow analysis in data value prediction
No full textValue prediction is a technique to increase parallelism by attempting to overcome serialization constraints caused by true data dependences. By predicting the outcome of an instruction before it executes, value prediction allows data dependent instructions to issue and execute speculatively, hence increasing parallelism when the prediction is correct. In case of a misprediction, the execution is redone with the corrected value. If the benefit from increased parallelism outweighs the misprediction recovery penalty, overall performance could be improved. Enhancing performance with value prediction therefore requires highly accurate prediction methods. Most existing general value prediction techniques are local and future outputs of an instruction are predicted based on outputs from previous executions of the same instruction. In this paper, we explore the possibility of introducing highly accurate general correlating value predictor using dynamic information flow analysis. We use information theory to mathematically prove the validity and benefits of correlating value predictors. We also introduce the concept of linear value predictors, a new technique that predicts a new value from another one using a linear relation. We then conduct empirical analysis using programs from SPECjvm2008 and Siemens benchmarks. Our empirical measurements support our mathematical theory and allow us to make important observations on the relation between predictability of data values and information flow. Furthermore, we provide a scheme to select highly predictable variables, and explain when a specific value predictor can perform well and even outperform other predictors. Using a dynamic information flow analysis tool called DynFlow, we show that the values of a set of selected variables can be predicted with very high accuracy, up to 100percent, from previous history of the same variables or other variables that have strong information flow into the predicted variable. © 2010 ACM.ACM SIGARCH;IEEE TCPP and TCCA;IFIP WG 10.3; Austrian Acad Sci ViennaAhuja P, 1998, P 12 INT C SUP JUL, P101, DOI 10.1145-277830.277854; AKKARY H, 1998, MICRO 31 NOV; AKKARY H, 2003, ICS 17, P12; AKKARY H, 2008, IFMT NOV; ALZAWAWI AS, 2007, ISCA 34 JUN; ASANOVI K, 2006, UCBEECS2006183; Austin T.M, 1992, P 19 ANN INT S COMP, V20, P342; BISHOP M, 2002, COMPUTER SECURITY AR, P2; BORKAR S, 2007, 44 ACM IEEE DES AUT, V4, P746; BURTSCHER M, 1999, INT C PAR ARCH COMP, P66; Butler M., 1991, P 18 INT S COMP ARCH, P276, DOI 10.1145-115952.115980; CHRYSOS G. Z, 1998, P 25 ANN INT S COMP; Cover TM, 1991, ELEMENTS INFORM THEO; Dalton M., 2007, P 34 INT S COMP ARCH; Denning D. E., 1982, CRYPTOGRAPHY DATA SE; EICKEMEYER RJ, 1993, IBM J RES DEV, V37, P547; Ellis JR, 1986, BULLDOG COMPILER VLI; Fano R. M., 1961, TRANSMISSION INFORM; FRANKLIN M, 1993, THESIS U WISCONSIN; GANDHI A., 2004, P 10 INT S HIGH PERF; GOEMAN B, P 7 INT S HIGH PERF; Harrold MJ, SIEMENS PROGRAMS HR; Hu Shiwen, 2003, J INSTRUCTION LEVEL, V5, P1; HWU WW, 1995, P IEEE, V83; Kachigan S. K., 1986, STAT ANAL INTERDISCI; KRUS D. J, 2006, VISUAL STAT; Kwak JW, 2007, MICROPROCESS MICROSY, V31, P63, DOI 10.1016-j.micpro.2006.08.002; LAM MS, 1992, ACM COMP AR, V20, P46, DOI 10.1109-ISCA.1992.753303; Lipasti M. H., 1996, P 7 INT C ARCH SUPP, P138, DOI 10.1145-237090.237173; Lipasti MH, 1996, PROCEEDINGS OF THE 29TH ANNUAL IEEE-ACM INTERNATIONAL SYMPOSIUM ON MICROARCHITECTURE - MICRO-29, P226, DOI 10.1109-MICRO.1996.566464; Marcuello P, 2004, IEEE T COMPUT, V53, P114, DOI 10.1109-TC.2004.1261823; MARCUELLO P, 1999, P 32 INT S MICR NOV; Martin R. P., 1997, SIGARCH COMPUTER ARC, V25, P85; MASRI W, 2006, WODA, P73; Masri W, 2009, INFORM SOFTWARE TECH, V51, P385, DOI 10.1016-j.infsof.2008.05.008; MASRI W, 2004, 15 IEEE INT S SOFTW; Masri W, 2009, ACM T SOFTW ENG METH, V19, DOI 10.1145-1571629.1571631; MOHAN W, 1992, P 5 INT C ARCH SUPP, P76; MONTOYE RK, 1990, IBM J RES DEV, V34, P59; Moshovos A, 2002, IEEE T COMPUT, V51, P313, DOI 10.1109-12.990129; MOSHOVOS A., 1997, P 24 ANN ACM IEEE C; MOSHOVOS A. I, 1998, THESIS U WISCONSIN M; Nakra T., 1999, Proceedings Fifth International Symposium on High-Performance Computer Architecture, DOI 10.1109-HPCA.1999.744311; OPLINGER J., 1999, P 8 INT C PAR ARCH C; Owens JD, 2007, IEEE MICRO, V27, P96, DOI 10.1109-MM.2007.4378787; PICKETT C. J. F, 2004, SABLETR20046 MCGILL; Pickett C.J.F., 2004, P 2 VAL PRED VAL BAS, P40; RYCHLIK B, 1998, 1998 INT C PAR ARCH; SAZEIDES Y, 1997, ECETR978 U WISC; SAZEIDES Y, 1996, MICRO 29 DEC; SAZEIDES Y, 1997, P 30 INT S MICR DEC; Siegel S., 1956, NONPARAMETRIC STAT B; SINGER J, 2006, QAPL 06, V164, P137; Smith A, 2006, INT SYMP MICROARCH, P89; Smith JE, 1995, P IEEE, V83, P1609, DOI 10.1109-5.476078; SODANI A, 1998, MICRO 31 P 31 ANN AC, P205; SODANI A, 1997, P 24 INT S COMP ARCH; *SPEC, SPEC BENCHM SUIT; STEFFAN J., 1999, P 8 INT S HIGH PERF; Thomas R, 2003, CONF PROC INT SYMP C, P314, DOI 10.1109-ISCA.2003.1207010; Tuck N., 2005, P 11 INT S HIGH PERF; TULLSEN D, 1999, 26 INT S COMP ARCH M; Venkataramani G., 2008, IEEE 14 INT S HIGH P, P173; Wang K, 1997, INT SYMP MICROARCH, P281; WARTER NJ, 1993, P 26 HICSS JAN, V1, P49734
