No abstract

Most existing person re-identification (Re-ID) approaches follow a supervised learning framework, in which a large number of labelled matching pairs are required for training. This severely limits their scalability in realworld applications. To overcome this limitation, we develop a novel cross-dataset transfer learning approach to learn a discriminative representation. It is unsupervised in the sense that the target dataset is completely unlabelled. Specifically, we present an multi-task dictionary learning method which is able to learn a dataset-shared but targetdata-biased representation. Experimental results on five benchmark datasets demonstrate that the method significantly outperforms the state-of-the-art.

No abstract

Person re-identification (Re-ID) poses a unique challenge to deep learning: how to learn a deep model with millions of parameters on a small training set of few or no labels. In this paper, a number of deep transfer learning models are proposed to address the data sparsity problem. First, a deep network architecture is designed which differs from existing deep Re-ID models in that (a) it is more suitable for transferring representations learned from large image classification datasets, and (b) classification loss and verification loss are combined, each of which adopts a different dropout strategy. Second, a two-stepped fine-tuning strategy is developed to transfer knowledge from auxiliary datasets. Third, given an unlabelled Re-ID dataset, a novel unsupervised deep transfer learning model is developed based on co-training. The proposed models outperform the state-ofthe-art deep Re-ID models by large margins: we achieve Rank-1 accuracy of 85.4%, 83.7% and 56.3% on CUHK03, Market1501, and VIPeR respectively, whilst on VIPeR, our unsupervised model (45.1%) beats most supervised models.

Every year more than 500,000 deaths are attributed to trauma worldwide and severe hemorrhage is present in most of them. Transfused platelets have been shown to improve survival in trauma patients, although its mechanism is only partially known. Platelet derived-extracellular vesicles (PEVs) are small vesicles released from platelets upon activation and/or mechanical stimulation and many of the benefits attributed to platelets could be mediated through PEVs. Based on the available literature, we hypothesized that transfusion of human PEVs would promote hemostasis, reduce blood loss and attenuate the progression to hemorrhagic shock following severe trauma. In this study, platelet units from four different donors were centrifuged to separate platelets and PEVs. The pellets were washed to obtain plasma-free platelets to use in the rodent model. The supernatant was subjected to tangential flow filtration for isolation and purification of PEVs. PEVs were assessed by total count and particle size distribution by Nanoparticle Tracking Analysis (NTA) and characterized for cells of origin and expression of EV specific-surface and cytosolic markers by flow cytometry. The coagulation profile from PEVs was assessed by calibrated automated thrombography (CAT) and thromboelastography (TEG). A rat model of uncontrolled hemorrhage was used to compare the therapeutic effects of 8.7 × 108 fresh platelets (FPLT group, n = 8), 7.8 × 109 PEVs (PEV group, n = 8) or Vehicle (Control, n = 16) following severe trauma. The obtained pool of PEVs from 4 donors had a mean size of 101 ± 47 nm and expressed the platelet-specific surface marker CD41 and the EV specific markers CD9, CD61, CD63, CD81 and HSP90. All PEV isolates demonstrated a dose-dependent increase in the rate and amount of thrombin generated and overall clot strength. In vivo experiments demonstrated a 24% reduction in abdominal blood loss following liver trauma in the PEVs group when compared with the control group (9.9 ± 0.4 vs. 7.5 ± 0.5 mL, p < 0.001>). The PEV group also exhibited improved outcomes in blood pressure, lactate level, base excess and plasma protein concentration compared to the Control group. Fresh platelets failed to improve these endpoints when compared to Controls. Altogether, these results indicate that human PEVs provide pro-hemostatic support following uncontrolled bleeding. As an additional therapeutic effect, PEVs improve the outcome following severe trauma by maintaining hemodynamic stability and attenuating the development of ischemia, base deficit, and cardiovascular shock.

Few-shot learning, which aims at extracting new concepts rapidly from extremely few examples of novel classes, has been featured into the meta-learning paradigm recently. Yet, the key challenge of how to learn a generalizable classifier with the capability of adapting to specific tasks with severely limited data still remains in this domain. To this end, we propose a Transductive Episodic-wise Adaptive Metric (TEAM) framework for few-shot learning, by integrating the meta-learning paradigm with both deep metric learning and transductive inference. With exploring the pairwise constraints and regularization prior within each task, we explicitly formulate the adaptation procedure into a standard semi-definite programming problem. By solving the problem with its closed-form solution on the fly with the setup of transduction, our approach efficiently tailors an episodic-wise metric for each task to adapt all features from a shared task-agnostic embedding space into a more discriminative task-specific metric space. Moreover, we further leverage an attention-based bi-directional similarity strategy for extracting the more robust relationship between queries and prototypes. Extensive experiments on three benchmark datasets show that our framework is superior to other existing approaches and achieves the state-ofthe-art performance in the few-shot literature.

No abstract

Abstract-Learning the spatial-temporal representation of motion information is crucial to human action recognition. Nevertheless, most of the existing features or descriptors cannot capture motion information effectively, especially for long-term motion. To address this problem, this paper proposes a long-term motion descriptor called sequential Deep Trajectory Descriptor (sDTD). Specifically, we project dense trajectories into twodimensional planes, and subsequently a CNN-RNN network is employed to learn an effective representation for long-term motion. Unlike the popular two-stream ConvNets, the sDTD stream is introduced into a three-stream framework so as to identify actions from a video sequence. Consequently, this threestream framework can simultaneously capture static spatial features, short-term motion and long-term motion in the video. Extensive experiments were conducted on three challenging datasets: KTH, HMDB51 and UCF101. Experimental results show that our method achieves state-of-the-art performance on the KTH and UCF101 datasets, and is comparable to the stateof-the-art methods on the HMDB51 dataset.