1. |
|
|
2. |
|
|
3. |
|
|
4. |
- Bravo, L, et al.
(författare)
-
- 2021
-
swepub:Mat__t
|
|
5. |
- Tabiri, S, et al.
(författare)
-
- 2021
-
swepub:Mat__t
|
|
6. |
- Glasbey, JC, et al.
(författare)
-
- 2021
-
swepub:Mat__t
|
|
7. |
- Axfors, Cathrine, et al.
(författare)
-
Association between convalescent plasma treatment and mortality in COVID-19 : a collaborative systematic review and meta-analysis of randomized clinical trials
- 2021
-
Ingår i: BMC Infectious Diseases. - : BioMed Central (BMC). - 1471-2334. ; 21:1
-
Forskningsöversikt (refereegranskat)abstract
- Background: Convalescent plasma has been widely used to treat COVID-19 and is under investigation in numerous randomized clinical trials, but results are publicly available only for a small number of trials. The objective of this study was to assess the benefits of convalescent plasma treatment compared to placebo or no treatment and all-cause mortality in patients with COVID-19, using data from all available randomized clinical trials, including unpublished and ongoing trials (Open Science Framework, ). Methods: In this collaborative systematic review and meta-analysis, clinical trial registries (ClinicalTrials.gov, WHO International Clinical Trials Registry Platform), the Cochrane COVID-19 register, the LOVE database, and PubMed were searched until April 8, 2021. Investigators of trials registered by March 1, 2021, without published results were contacted via email. Eligible were ongoing, discontinued and completed randomized clinical trials that compared convalescent plasma with placebo or no treatment in COVID-19 patients, regardless of setting or treatment schedule. Aggregated mortality data were extracted from publications or provided by investigators of unpublished trials and combined using the Hartung-Knapp-Sidik-Jonkman random effects model. We investigated the contribution of unpublished trials to the overall evidence. Results: A total of 16,477 patients were included in 33 trials (20 unpublished with 3190 patients, 13 published with 13,287 patients). 32 trials enrolled only hospitalized patients (including 3 with only intensive care unit patients). Risk of bias was low for 29/33 trials. Of 8495 patients who received convalescent plasma, 1997 died (23%), and of 7982 control patients, 1952 died (24%). The combined risk ratio for all-cause mortality was 0.97 (95% confidence interval: 0.92; 1.02) with between-study heterogeneity not beyond chance (I-2 = 0%). The RECOVERY trial had 69.8% and the unpublished evidence 25.3% of the weight in the meta-analysis. Conclusions: Convalescent plasma treatment of patients with COVID-19 did not reduce all-cause mortality. These results provide strong evidence that convalescent plasma treatment for patients with COVID-19 should not be used outside of randomized trials. Evidence synthesis from collaborations among trial investigators can inform both evidence generation and evidence application in patient care.
|
|
8. |
- Aamodt, K., et al.
(författare)
-
Alignment of the ALICE Inner Tracking System with cosmic-ray tracks
- 2010
-
Ingår i: Journal of Instrumentation. - 1748-0221. ; 5
-
Konferensbidrag (refereegranskat)abstract
- ALICE (A Large Ion Collider Experiment) is the LHC (Large Hadron Collider) experiment devoted to investigating the strongly interacting matter created in nucleus-nucleus collisions at the LHC energies. The ALICE ITS, Inner Tracking System, consists of six cylindrical layers of silicon detectors with three different technologies; in the outward direction: two layers of pixel detectors, two layers each of drift, and strip detectors. The number of parameters to be determined in the spatial alignment of the 2198 sensor modules of the ITS is about 13,000. The target alignment precision is well below 10 mu m in some cases (pixels). The sources of alignment information include survey measurements, and the reconstructed tracks from cosmic rays and from proton-proton collisions. The main track-based alignment method uses the Millepede global approach. An iterative local method was developed and used as well. We present the results obtained for the ITS alignment using about 10(5) charged tracks from cosmic rays that have been collected during summer 2008, with the ALICE solenoidal magnet switched off.
|
|
9. |
- Aamodt, K., et al.
(författare)
-
Charged-particle multiplicity measurement in proton-proton collisions at root s=0.9 and 2.36 TeV with ALICE at LHC
- 2010
-
Ingår i: European Physical Journal C. Particles and Fields. - : Springer Science and Business Media LLC. - 1434-6044. ; 68:1-2, s. 89-108
-
Tidskriftsartikel (refereegranskat)abstract
- Charged-particle production was studied in proton-proton collisions collected at the LHC with the ALICE detector at centre-of-mass energies 0.9 TeV and 2.36 TeV in the pseudorapidity range vertical bar eta vertical bar < 1.4. In the central region (vertical bar eta vertical bar < 0.5), at 0.9 TeV, we measure charged-particle pseudo-rapidity density dN(ch)/d eta = 3.02 +/- 0.01(stat.)(-0.05)(+0.08)(syst.) for inelastic interactions, and dN(ch)/d eta = 3.58 +/- 0.01 (stat.)(-0.12)(+0.12)(syst.) for non-single-diffractive interactions. At 2.36 TeV, we find dN(ch)/d eta = 3.77 +/- 0.01(stat.)(-0.12)(+0.25)(syst.) for inelastic, and dN(ch)/d eta = 4.43 +/- 0.01(stat.)(-0.12)(+0.17)(syst.) for non-single-diffractive collisions. The relative increase in charged-particle multiplicity from the lower to higher energy is 24.7% +/- 0.5%(stat.)(-2.8)(+5.7)%(syst.) for inelastic and 23.7% +/- 0.5%(stat.)(-1.1)(+4.6)%(syst.) for non-single-diffractive interactions. This increase is consistent with that reported by the CMS collaboration for non-single-diffractive events and larger than that found by a number of commonly used models. The multiplicity distribution was measured in different pseudorapidity intervals and studied in terms of KNO variables at both energies. The results are compared to proton-antiproton data and to model predictions.
|
|
10. |
- Aamodt, K., et al.
(författare)
-
Charged-particle multiplicity measurement in proton-proton collisions at root s=7 TeV with ALICE at LHC
- 2010
-
Ingår i: European Physical Journal C. Particles and Fields. - : Springer Science and Business Media LLC. - 1434-6044. ; 68:3-4, s. 345-354
-
Tidskriftsartikel (refereegranskat)abstract
- The pseudorapidity density and multiplicity distribution of charged particles produced in proton-proton collisions at the LHC, at a centre-of-mass energy root s = 7 TeV, were measured in the central pseudorapidity region vertical bar eta vertical bar < 1. Comparisons are made with previous measurements at root s = 0.9 TeV and 2.36 TeV. At root s = 7 TeV, for events with at least one charged particle in |eta vertical bar| < 1, we obtain dN(ch)/d eta = 6.01 +/- 0.01(stat.)(-0.12)(+0.20) (syst.). This corresponds to an increase of 57.6%+/-0.4%(stat.)(-1.8%)(+3.6) (syst.) relative to collisions at 0.9 TeV, significantly higher than calculations from commonly used models. The multiplicity distribution at 7 TeV is described fairly well by the negative binomial distribution.
|
|
11. |
- Aamodt, K., et al.
(författare)
-
First proton-proton collisions at the LHC as observed with the ALICE detector: measurement of the charged-particle pseudorapidity density at root s=900 GeV
- 2010
-
Ingår i: European Physical Journal C. Particles and Fields. - : Springer Science and Business Media LLC. - 1434-6044. ; 65:1-2, s. 111-125
-
Tidskriftsartikel (refereegranskat)abstract
- On 23rd November 2009, during the early commissioning of the CERN Large Hadron Collider (LHC), two counter-rotating proton bunches were circulated for the first time concurrently in the machine, at the LHC injection energy of 450 GeV per beam. Although the proton intensity was very low, with only one pilot bunch per beam, and no systematic attempt was made to optimize the collision optics, all LHC experiments reported a number of collision candidates. In the ALICE experiment, the collision region was centred very well in both the longitudinal and transverse directions and 284 events were recorded in coincidence with the two passing proton bunches. The events were immediately reconstructed and analyzed both online and offline. We have used these events to measure the pseudorapidity density of charged primary particles in the central region. In the range vertical bar eta vertical bar < 0.5, we obtain dN(ch)/d eta = 3.10 +/- 0.13(stat.) +/- 0.22(syst.) for all inelastic interactions, and dN(ch)/d eta = 3.51 +/- 0.15(stat.) +/- 0.25(syst.) for nonsingle diffractive interactions. These results are consistent with previous measurements in proton-antiproton interactions at the same centre-of-mass energy at the CERN Sp<(p)over bar>S collider. They also illustrate the excellent functioning and rapid progress of the LHC accelerator, and of both the hardware and software of the ALICE experiment, in this early start-up phase.
|
|
12. |
- Aamodt, K., et al.
(författare)
-
Midrapidity Antiproton-to-Proton Ratio in pp Collisons root s=0.9 and 7 TeV Measured by the ALICE Experiment
- 2010
-
Ingår i: Physical Review Letters. - 1079-7114. ; 105:7
-
Tidskriftsartikel (refereegranskat)abstract
- The ratio of the yields of antiprotons to protons in pp collisions has been measured by the ALICE experiment at root s = 0.9 and 7 TeV during the initial running periods of the Large Hadron Collider. The measurement covers the transverse momentum interval 0.45 < p(t) < 1.05 GeV/c and rapidity vertical bar y vertical bar < 0.5. The ratio is measured to be R-vertical bar y vertical bar<0.5 = 0.957 +/- 0.006(stat) +/- 0.0014(syst) at 0.9 Tev and R-vertical bar y vertical bar<0.5 = 0.991 +/- 0.005 +/- 0.014(syst) at 7 TeV and it is independent of both rapidity and transverse momentum. The results are consistent with the conventional model of baryon-number transport and set stringent limits on any additional contributions to baryon-number transfer over very large rapidity intervals in pp collisions.
|
|
13. |
- Aamodt, K., et al.
(författare)
-
Production of pions, kaons and protons in pp collisions at root s=900 GeV with ALICE at the LHC
- 2011
-
Ingår i: European Physical Journal C. Particles and Fields. - : Springer Science and Business Media LLC. - 1434-6044. ; 71:6
-
Tidskriftsartikel (refereegranskat)abstract
- The production of pi(+), pi(-), K+, K-, p, and (p) over bar at mid-rapidity has been measured in proton-proton collisions at root s = 900 GeV with the ALICE detector. Particle identification is performed using the specific energy loss in the inner tracking silicon detector and the time projection chamber. In addition, time-of-flight information is used to identify hadrons at higher momenta. Finally, the distinctive kink topology of the weak decay of charged kaons is used for an alternative measurement of the kaon transverse momentum (p(t)) spectra. Since these various particle identification tools give the best separation capabilities over different momentum ranges, the results are combined to extract spectra from p(t) = 100 MeV/c to 2.5 GeV/c. The measured spectra are further compared with QCD-inspired models which yield a poor description. The total yields and the mean pt are compared with previous measurements, and the trends as a function of collision energy are discussed.
|
|
14. |
- Aamodt, K., et al.
(författare)
-
Transverse momentum spectra of charged particles in proton-proton collisions at root s=900 GeV with ALICE at the LHC
- 2010
-
Ingår i: Physics Letters. Section B: Nuclear, Elementary Particle and High-Energy Physics. - : Elsevier BV. - 0370-2693. ; 693:2, s. 53-68
-
Tidskriftsartikel (refereegranskat)abstract
- The inclusive charged particle transverse momentum distribution is measured in proton-proton collisions at root s = 900 GeV at the LHC using the ALICE detector. The measurement is performed in the central pseudorapidity region (vertical bar eta vertical bar < 0.8) over the transverse momentum range 0.15 < p(T) < 10 GeV/c. The correlation between transverse momentum and particle multiplicity is also studied. Results are presented for inelastic (INEL) and non-single-diffractive (NSD) events. The average transverse momentum for vertical bar eta vertical bar < 0.8 is < p(T)>(INEL) = 0.483 +/- 0.001 (stat.) +/- 0.007 (syst.) GeV/c and < p(T)>(NSD) = 0.489 +/- 0.001 (stat.) +/- 0.007 (syst.) GeV/c, respectively. The data exhibit a slightly larger < p(T)> than measurements in wider pseudorapidity intervals. The results are compared to simulations with the Monte Carlo event generators PYTHIA and PHOJET. (C) 2010 Published by Elsevier B.V.
|
|
15. |
- Aamodt, K., et al.
(författare)
-
Two-pion Bose-Einstein correlations in pp collisions at root s=900 GeV
- 2010
-
Ingår i: Physical Review D (Particles, Fields, Gravitation and Cosmology). - 1550-2368. ; 82:5
-
Tidskriftsartikel (refereegranskat)abstract
- We report on the measurement of two-pion correlation functions from pp collisions at root s = 900 GeV performed by the ALICE experiment at the Large Hadron Collider. Our analysis shows an increase of the Hanbury Brown-Twiss radius with increasing event multiplicity, in line with other measurements done in particle- and nuclear collisions. Conversely, the strong decrease of the radius with increasing transverse momentum, as observed at the Relativistic Heavy Ion Collider and at Tevatron, is not manifest in our data.
|
|
16. |
- Kanai, M, et al.
(författare)
-
- 2023
-
swepub:Mat__t
|
|
17. |
- Kristan, Matej, et al.
(författare)
-
The Sixth Visual Object Tracking VOT2018 Challenge Results
- 2019
-
Ingår i: Computer Vision – ECCV 2018 Workshops. - Cham : Springer Publishing Company. - 9783030110086 - 9783030110093 ; , s. 3-53
-
Konferensbidrag (refereegranskat)abstract
- The Visual Object Tracking challenge VOT2018 is the sixth annual tracker benchmarking activity organized by the VOT initiative. Results of over eighty trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years. The evaluation included the standard VOT and other popular methodologies for short-term tracking analysis and a “real-time” experiment simulating a situation where a tracker processes images as if provided by a continuously running sensor. A long-term tracking subchallenge has been introduced to the set of standard VOT sub-challenges. The new subchallenge focuses on long-term tracking properties, namely coping with target disappearance and reappearance. A new dataset has been compiled and a performance evaluation methodology that focuses on long-term tracking capabilities has been adopted. The VOT toolkit has been updated to support both standard short-term and the new long-term tracking subchallenges. Performance of the tested trackers typically by far exceeds standard baselines. The source code for most of the trackers is publicly available from the VOT page. The dataset, the evaluation kit and the results are publicly available at the challenge website (http://votchallenge.net).
|
|
18. |
- Rodríguez-García, Carlota, et al.
(författare)
-
Efficacy of mepolizumab in usual clinical practice and characteristics of responders : Mepolizumab in usual clinical practice
- 2021
-
Ingår i: Respiratory Medicine. - : Elsevier BV. - 0954-6111. ; 187
-
Tidskriftsartikel (refereegranskat)abstract
- Background: Severe eosinophilic asthma is a high-burden disease. Mepolizumab has been effective in several randomized clinical trials. However, such success might not be applicable to patients treated in usual clinical practice. The objectives of this article are to evaluate the efficacy of mepolizumab in severe uncontrolled eosinophilic asthma under usual clinical practice, and to determine characteristics associated with the response to this treatment. Methods: We have conducted a retrospective, multicentre study, including all adult patients with severe uncontrolled eosinophilic asthma in Galicia, Spain, on whom mepolizumab treatment was started before June 2020, at least 6 months before the time of inclusion, and had received at least one dose of the drug. Patient characteristics, clinical data, respiratory function and comorbidities were collected at baseline and at the 6-month-follow-up. Responders and super-responders were defined according to clinical response and requirement of systemic corticosteroids. Results: 122 patients (mean age 58 years old) were included. In the follow-up treatment 6 months later, 75.4% of the patients were well-controlled, displaying a significant reduction in blood eosinophil counts (p < 0.001), hospital admissions and disease exacerbations (p < 0.001), and had their systemic glucocorticosteroid dose significantly reduced (p < 0.001). The inhaled corticosteroid dose was also lowered (p < 0.01) after 6 months of treatment. Around two-thirds had a clinically significant increase in FEV1, 95% of the patients were considered responders and 43% super-responders. Conclusion: In routine clinical practice, mepolizumab is effective in patients with severe eosinophilic asthma and it has a good safety profile.
|
|
19. |
- Zhang, Q, et al.
(författare)
-
Autoantibodies against type I IFNs in patients with critical influenza pneumonia
- 2022
-
Ingår i: The Journal of experimental medicine. - : Rockefeller University Press. - 1540-9538 .- 0022-1007. ; 219:11
-
Tidskriftsartikel (refereegranskat)abstract
- Autoantibodies neutralizing type I interferons (IFNs) can underlie critical COVID-19 pneumonia and yellow fever vaccine disease. We report here on 13 patients harboring autoantibodies neutralizing IFN-α2 alone (five patients) or with IFN-ω (eight patients) from a cohort of 279 patients (4.7%) aged 6–73 yr with critical influenza pneumonia. Nine and four patients had antibodies neutralizing high and low concentrations, respectively, of IFN-α2, and six and two patients had antibodies neutralizing high and low concentrations, respectively, of IFN-ω. The patients’ autoantibodies increased influenza A virus replication in both A549 cells and reconstituted human airway epithelia. The prevalence of these antibodies was significantly higher than that in the general population for patients <70 yr of age (5.7 vs. 1.1%, P = 2.2 × 10−5), but not >70 yr of age (3.1 vs. 4.4%, P = 0.68). The risk of critical influenza was highest in patients with antibodies neutralizing high concentrations of both IFN-α2 and IFN-ω (OR = 11.7, P = 1.3 × 10−5), especially those <70 yr old (OR = 139.9, P = 3.1 × 10−10). We also identified 10 patients in additional influenza patient cohorts. Autoantibodies neutralizing type I IFNs account for ∼5% of cases of life-threatening influenza pneumonia in patients <70 yr old.
|
|
20. |
|
|
21. |
- El Maatougui, Abdelaziz, et al.
(författare)
-
Discovery of Potent and Highly Selective A(2B) Adenosine Receptor Antagonist Chemotypes
- 2016
-
Ingår i: Journal of Medicinal Chemistry. - : American Chemical Society (ACS). - 0022-2623 .- 1520-4804. ; 59:5, s. 1967-1983
-
Tidskriftsartikel (refereegranskat)abstract
- Three novel families of A(2B) adenosine receptor antagonists were identified in the context of the structural exploration of the 3,4-dihydropyrimidin,2(1H)-one chemotype. The most appealing series contain imidazole, 1,2,4-triazole, or benzimidazole rings fused to the 2,3-positions of the parent diazinone core. The optimization process enabled identification of a highly potent (3.49 nM) A(2B) ligand that exhibits complete selectivity toward A(1), A(2A), and A(3) receptors. The results of functional cAMP experiments confirmed the antagonistic behavior of representative ligands. The main SAR trends identified within the series were substantiated by a molecular modeling study based on a receptor-driven docking model constructed on the basis of the crystal structure of the human A(2A) receptor.
|
|
22. |
- García, Diego, et al.
(författare)
-
Conditioned Fully Convolutional Denoising Autoencoder for Energy Disaggregation
- 2023
-
Ingår i: Artificial Intelligence Applications and Innovations. AIAI 2023 IFIP WG 12.5 International Workshops. - : Springer Nature. - 9783031341731 - 9783031341717 ; , s. 421-433
-
Konferensbidrag (refereegranskat)abstract
- Energy management increasingly requires tools to support decisions for improving consumption. This is achieved not only obtaining feedback from current systems but also using prior knowledge about human behaviour. The advances of data-driven models provide techniques like Non-Intrusive Load Monitoring (NILM) which are capable of estimating energy demand of appliances from total consumption. In addition, deep learning models have improved accuracy in energy disaggregation using separated networks for each device. However, the complexity can increase in large facilities and feedback may be impaired for a proper interpretation. In this work, a deep neural network based on a Fully Convolutional denoising AutoEncoder is proposed for energy disaggregation that uses a conditioning input to modulate the estimation aimed to one specific appliance. The model performs a complete disaggregation using a network whose modulation to target the estimation can be steered by the user. Experiments are done using data from a hospital facility and evaluating reconstruction errors and computational efficiency. The results show acceptable errors compared to methods that require various networks and a reduction of the complexity and computational costs, which can allow the user to be integrated into the analysis loop.
|
|
23. |
- Kristanl, Matej, et al.
(författare)
-
The Seventh Visual Object Tracking VOT2019 Challenge Results
- 2019
-
Ingår i: 2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW). - : IEEE COMPUTER SOC. - 9781728150239 ; , s. 2206-2241
-
Konferensbidrag (refereegranskat)abstract
- The Visual Object Tracking challenge VOT2019 is the seventh annual tracker benchmarking activity organized by the VOT initiative. Results of 81 trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years. The evaluation included the standard VOT and other popular methodologies for short-term tracking analysis as well as the standard VOT methodology for long-term tracking analysis. The VOT2019 challenge was composed of five challenges focusing on different tracking domains: (i) VOT-ST2019 challenge focused on short-term tracking in RGB, (ii) VOT-RT2019 challenge focused on "real-time" short-term tracking in RGB, (iii) VOT-LT2019 focused on long-term tracking namely coping with target disappearance and reappearance. Two new challenges have been introduced: (iv) VOT-RGBT2019 challenge focused on short-term tracking in RGB and thermal imagery and (v) VOT-RGBD2019 challenge focused on long-term tracking in RGB and depth imagery. The VOT-ST2019, VOT-RT2019 and VOT-LT2019 datasets were refreshed while new datasets were introduced for VOT-RGBT2019 and VOT-RGBD2019. The VOT toolkit has been updated to support both standard short-term, long-term tracking and tracking with multi-channel imagery. Performance of the tested trackers typically by far exceeds standard baselines. The source code for most of the trackers is publicly available from the VOT page. The dataset, the evaluation kit and the results are publicly available at the challenge website(1).
|
|
24. |
- Matheu, Victor, et al.
(författare)
-
Insulin allergy and resistance successfully treated by desensitisation with Aspart insulin
- 2005
-
Ingår i: Clinical and Molecular Allergy. - : Springer Science and Business Media LLC. - 1476-7961. ; 3:16
-
Tidskriftsartikel (refereegranskat)abstract
- A 25-year-old, with type I Diabetes Mellitus with a previous diagnosis of Protamine Allergy but not to human Insulin, started to notice anaphylactic reactions immediately after bolus with Insulin. Skin prick and intradermal test were positive to all insulins. Skin tests to other potential allergens resulted negative. Examination after bolus of Human Insulin revealed urticaria. Daily insulin requirement were around 2-2,4 U/Kg/day. Slow desensitisation with Aspart insulin, the insulin with lowest size of skin test, was performed using subcutaneous insulin pump. Six months after the end of desensitisation his daily insulin requirement decreased to 0.8 U/Kg/day and oral corticosteroids are being reduced with no symptoms.
|
|
25. |
- Serge, M. A., et al.
(författare)
-
Testing the Effect of Relative Pollen Productivity on the REVEALS Model : A Validated Reconstruction of Europe-Wide Holocene Vegetation
- 2023
-
Ingår i: Land. - : MDPI. - 2073-445X. ; 12:5
-
Tidskriftsartikel (refereegranskat)abstract
- Reliable quantitative vegetation reconstructions for Europe during the Holocene are crucial to improving our understanding of landscape dynamics, making it possible to assess the past effects of environmental variables and land-use change on ecosystems and biodiversity, and mitigating their effects in the future. We present here the most spatially extensive and temporally continuous pollen-based reconstructions of plant cover in Europe (at a spatial resolution of 1 degrees x 1 degrees) over the Holocene (last 11.7 ka BP) using the 'Regional Estimates of VEgetation Abundance from Large Sites' (REVEALS) model. This study has three main aims. First, to present the most accurate and reliable generation of REVEALS reconstructions across Europe so far. This has been achieved by including a larger number of pollen records compared to former analyses, in particular from the Mediterranean area. Second, to discuss methodological issues in the quantification of past land cover by using alternative datasets of relative pollen productivities (RPPs), one of the key input parameters of REVEALS, to test model sensitivity. Finally, to validate our reconstructions with the global forest change dataset. The results suggest that the RPPs.st1 (31 taxa) dataset is best suited to producing regional vegetation cover estimates for Europe. These reconstructions offer a long-term perspective providing unique possibilities to explore spatial-temporal changes in past land cover and biodiversity.
|
|
26. |
- Wang, Yaxing, et al.
(författare)
-
MineGAN plus plus : Mining Generative Models for Efficient Knowledge Transfer to Limited Data Domains
- 2023
-
Ingår i: International Journal of Computer Vision. - : SPRINGER. - 0920-5691 .- 1573-1405. ; 132:2, s. 490-514
-
Tidskriftsartikel (refereegranskat)abstract
- Given the often enormous effort required to train GANs, both computationally as well as in dataset collection, the re-use of pretrained GANs largely increases the potential impact of generative models. Therefore, we propose a novel knowledge transfer method for generative models based on mining the knowledge that is most beneficial to a specific target domain, either from a single or multiple pretrained GANs. This is done using a miner network that identifies which part of the generative distribution of each pretrained GAN outputs samples closest to the target domain. Mining effectively steers GAN sampling towards suitable regions of the latent space, which facilitates the posterior finetuning and avoids pathologies of other methods, such as mode collapse and lack of flexibility. Furthermore, to prevent overfitting on small target domains, we introduce sparse subnetwork selection, that restricts the set of trainable neurons to those that are relevant for the target dataset. We perform comprehensive experiments on several challenging datasets using various GAN architectures (BigGAN, Progressive GAN, and StyleGAN) and show that the proposed method, called MineGAN, effectively transfers knowledge to domains with few target images, outperforming existing methods. In addition, MineGAN can successfully transfer knowledge from multiple pretrained GANs. MineGAN.
|
|
27. |
- Zhang, Lichao, et al.
(författare)
-
Learning the Model Update for Siamese Trackers
- 2019
-
Ingår i: 2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019), Seoul, SOUTH KOREA, OCT 27-NOV 02, 2019. - : IEEE COMPUTER SOC. - 9781728148038 ; , s. 4009-4018
-
Konferensbidrag (refereegranskat)abstract
- Siamese approaches address the visual tracking problem by extracting an appearance template from the current frame, which is used to localize the target in the next frame. In general, this template is linearly combined with the accumulated template from the previous frame, resulting in an exponential decay of information over time. While such an approach to updating has led to improved results, its simplicity limits the potential gain likely to be obtained by learning to update. Therefore, we propose to replace the handcrafted update function with a method which learns to update. We use a convolutional neural network, called UpdateNet, which given the initial template, the accumulated template and the template of the current frame aims to estimate the optimal template for the next frame. The UpdateNet is compact and can easily be integrated into existing Siamese trackers. We demonstrate the generality of the proposed approach by applying it to two Siamese trackers, SiamFC and DaSiamRPN. Extensive experiments on VOT2016, VOT2018, LaSOT, and TrackingNet datasets demonstrate that our UpdateNet effectively predicts the new target template, outperforming the standard linear update. On the large-scale TrackingNet dataset, our UpdateNet improves the results of DaSiamRPN with an absolute gain of 3.9% in terms of success score. Code and models are available at https://github.com/zhanglichao/updatenet.
|
|
28. |
- Zhang, Lichao, et al.
(författare)
-
Synthetic Data Generation for End-to-End Thermal Infrared Tracking
- 2019
-
Ingår i: IEEE Transactions on Image Processing. - : IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC. - 1057-7149 .- 1941-0042. ; 28:4, s. 1837-1850
-
Tidskriftsartikel (refereegranskat)abstract
- The usage of both off-the-shelf and end-to-end trained deep networks have significantly improved the performance of visual tracking on RGB videos. However, the lack of large labeled datasets hampers the usage of convolutional neural networks for tracking in thermal infrared (TIR) images. Therefore, most state-of-the-art methods on tracking for TIR data are still based on handcrafted features. To address this problem, we propose to use image-to-image translation models. These models allow us to translate the abundantly available labeled RGB data to synthetic TIR data. We explore both the usage of paired and unpaired image translation models for this purpose. These methods provide us with a large labeled dataset of synthetic TIR sequences, on which we can train end-to-end optimal features for tracking. To the best of our knowledge, we are the first to train end-to-end features for TIR tracking. We perform extensive experiments on the VOT-TIR2017 dataset. We show that a network trained on a large dataset of synthetic TIR data obtains better performance than one trained on the available real TIR data. Combining both data sources leads to further improvement. In addition, when we combine the network with motion features, we outperform the state of the art with a relative gain of over 10%, clearly showing the efficiency of using synthetic data to train end-to-end TIR trackers.
|
|