arXiv subjects

Xin Jin

Publications and source records attributed to Xin Jin.

At least 397 records · Page 22Linked to original sources

arXiv2020

Style Normalization and Restitution for Generalizable Person Re-identification

Existing fully-supervised person re-identification (ReID) methods usually suffer from poor generalization capability caused by domain gaps. The key to solving this problem lies in filtering out identity-irrelevant interference and learning domain-invariant person representations. In this paper, we aim to design a generalizable person ReID framework which trains a model on source domains yet is able to generalize/perform well on target domains. To achieve this goal, we propose a simple yet effective Style Normalization and Restitution (SNR) module. Specifically, we filter out style variations (e.g., illumination, color contrast) by Instance Normalization (IN). However, such a process inevitably removes discriminative information. We propose to distill identity-relevant feature from the removed information and restitute it to the network to ensure high discrimination. For better disentanglement, we enforce a dual causal loss constraint in SNR to encourage the separation of identity-relevant features and identity-irrelevant features. Extensive experiments demonstrate the strong generalization capability of our framework. Our models empowered by the SNR modules significantly outperform the state-of-the-art domain generalization approaches on multiple widely-used person ReID benchmarks, and also show superiority on unsupervised domain adaptation.

cs.CV↗

arXiv2020

Learned Video Compression with Feature-level Residuals

In this paper, we present an end-to-end video compression network for P-frame challenge on CLIC. We focus on deep neural network (DNN) based video compression, and improve the current frameworks from three aspects. First, we notice that pixel space residuals is sensitive to the prediction errors of optical flow based motion compensation. To suppress the relative influence, we propose to compress the residuals of image feature rather than the residuals of image pixels. Furthermore, we combine the advantages of both pixel-level and feature-level residual compression methods by model ensembling. Finally, we propose a step-by-step training strategy to improve the training efficiency of the whole framework. Experiment results indicate that our proposed method achieves 0.9968 MS-SSIM on CLIC validation set and 0.9967 MS-SSIM on test set.

eess.IV↗

arXiv2020

Relation-Aware Global Attention for Person Re-identification

For person re-identification (re-id), attention mechanisms have become attractive as they aim at strengthening discriminative features and suppressing irrelevant ones, which matches well the key of re-id, i.e., discriminative feature learning. Previous approaches typically learn attention using local convolutions, ignoring the mining of knowledge from global structure patterns. Intuitively, the affinities among spatial positions/nodes in the feature map provide clustering-like information and are helpful for inferring semantics and thus attention, especially for person images where the feasible human poses are constrained. In this work, we propose an effective Relation-Aware Global Attention (RGA) module which captures the global structural information for better attention learning. Specifically, for each feature position, in order to compactly grasp the structural information of global scope and local appearance information, we propose to stack the relations, i.e., its pairwise correlations/affinities with all the feature positions (e.g., in raster scan order), and the feature itself together to learn the attention with a shallow convolutional model. Extensive ablation studies demonstrate that our RGA can significantly enhance the feature representation power and help achieve the state-of-the-art performance on several popular benchmarks. The source code is available at https://github.com/microsoft/Relation-Aware-Global-Attention-Networks.

cs.CV↗

arXiv2020

Exploring Categorical Regularization for Domain Adaptive Object Detection

In this paper, we tackle the domain adaptive object detection problem, where the main challenge lies in significant domain gaps between source and target domains. Previous work seeks to plainly align image-level and instance-level shifts to eventually minimize the domain discrepancy. However, they still overlook to match crucial image regions and important instances across domains, which will strongly affect domain shift mitigation. In this work, we propose a simple but effective categorical regularization framework for alleviating this issue. It can be applied as a plug-and-play component on a series of Domain Adaptive Faster R-CNN methods which are prominent for dealing with domain adaptive detection. Specifically, by integrating an image-level multi-label classifier upon the detection backbone, we can obtain the sparse but crucial image regions corresponding to categorical information, thanks to the weakly localization ability of the classification manner. Meanwhile, at the instance level, we leverage the categorical consistency between image-level predictions (by the classifier) and instance-level predictions (by the detection head) as a regularization factor to automatically hunt for the hard aligned instances of target domains. Extensive experiments of various domain shift scenarios show that our method obtains a significant performance gain over original Domain Adaptive Faster R-CNN detectors. Furthermore, qualitative visualization and analyses can demonstrate the ability of our method for attending on the key regions/instances targeting on domain adaptation. Our code is open-source and available at \url{https://github.com/Megvii-Nanjing/CR-DA-DET}.

cs.CV↗

arXiv2020

Semantics-Aligned Representation Learning for Person Re-identification

Person re-identification (reID) aims to match person images to retrieve the ones with the same identity. This is a challenging task, as the images to be matched are generally semantically misaligned due to the diversity of human poses and capture viewpoints, incompleteness of the visible bodies (due to occlusion), etc. In this paper, we propose a framework that drives the reID network to learn semantics-aligned feature representation through delicate supervision designs. Specifically, we build a Semantics Aligning Network (SAN) which consists of a base network as encoder (SA-Enc) for re-ID, and a decoder (SA-Dec) for reconstructing/regressing the densely semantics aligned full texture image. We jointly train the SAN under the supervisions of person re-identification and aligned texture generation. Moreover, at the decoder, besides the reconstruction loss, we add Triplet ReID constraints over the feature maps as the perceptual losses. The decoder is discarded in the inference and thus our scheme is computationally efficient. Ablation studies demonstrate the effectiveness of our design. We achieve the state-of-the-art performances on the benchmark datasets CUHK03, Market1501, MSMT17, and the partial person reID dataset Partial REID. Code for our proposed method is available at: https://github.com/microsoft/Semantics-Aligned-Representation-Learning-for-Person-Re-identification.

cs.CV↗

arXiv2020

Comparison of neutron and ion irradiation induced lattice parameter changes in Ni, Cu and MgO single crystals

Data available in the literature of lattice parameter changes in Ni, Cu and MgO single crystals irradiated in neutron reactors are tentatively reproduced using ad hoc ion irradiation experiments. The nature and energy of the ions were selected so that their weighted average recoil spectrum matches at best that of the neutron flux inside the reactor. The ion fluence was calculated to obtain similar, average displacements per atom (dpa) levels. The lattice parameters of the irradiated samples were determined using X-ray diffraction. We show that, for the metals, it is not possible to reproduce the neutron data, essentially because the dpa levels are low and the irradiated, damaged volume is small. On the contrary, for MgO, a very close lattice parameter change is obtained, providing that the irradiation temperature is significantly increased (1000 K) with respect to that in the reactor (300 K).

cond-mat.mtrl-sci↗

arXiv2020

Uncertainty-Aware Multi-Shot Knowledge Distillation for Image-Based Object Re-Identification

Object re-identification (re-id) aims to identify a specific object across times or camera views, with the person re-id and vehicle re-id as the most widely studied applications. Re-id is challenging because of the variations in viewpoints, (human) poses, and occlusions. Multi-shots of the same object can cover diverse viewpoints/poses and thus provide more comprehensive information. In this paper, we propose exploiting the multi-shots of the same identity to guide the feature learning of each individual image. Specifically, we design an Uncertainty-aware Multi-shot Teacher-Student (UMTS) Network. It consists of a teacher network (T-net) that learns the comprehensive features from multiple images of the same object, and a student network (S-net) that takes a single image as input. In particular, we take into account the data dependent heteroscedastic uncertainty for effectively transferring the knowledge from the T-net to S-net. To the best of our knowledge, we are the first to make use of multi-shots of an object in a teacher-student learning manner for effectively boosting the single image based re-id. We validate the effectiveness of our approach on the popular vehicle re-id and person re-id datasets. In inference, the S-net alone significantly outperforms the baselines and achieves the state-of-the-art performance.

cs.CV↗

arXiv2019

A New Strategy for Estimating Photometric Redshifts of Quasars

Based on the SDSS and SDSS-WISE quasar datasets, we put forward two schemes to estimate the photometric redshifts of quasars. Our schemes are based on the idea that the samples are firstly classified into subsamples by a classifier and then photometric redshift estimation of different subsamples is performed by a regressor. Random Forest is adopted as the core algorithm of the classifiers, while Random Forest and kNN are applied as the key algorithms of regressors. The samples are divided into two subsamples and four subsamples depending on the redshift distribution. The performance based on different samples, different algorithms and different schemes are compared. The experimental results indicate that the accuracy of photometric redshift estimation for the two schemes generally improve to some extent compared to the original scheme in terms of the percents in \frac{|Δz|}{1+z_{i}}<0.1 and \frac{|Δz|}{1+z_{i}}<0.2 and mean absolute error. Only given the running speed, kNN shows its superiority to Random Forest. The performance of Random Forest is a little better than or comparable to that of kNN with the two datasets. The accuracy based on the SDSS-WISE sample outperforms that based on the SDSS sample no matter by kNN or by Random Forest. More information from more bands is considered and helpful to improve the accuracy of photometric redshift estimation. Evidently it can be found that our strategy to estimate photometric redshift is applicable and may be applied to other datasets or other kinds of objects. Only talking about the percent in \frac{|Δz|}{1+z_{i}}<0.3, there is still large room for further improvement in the photometric redshift estimation.

astro-ph.IM↗

arXiv2019

Freeing electrons from extrinsic and intrinsic disorder yields band-like transport in n-type organic semiconductors

Charge transport in organic semiconductors is notoriously extremely sensitive to the presence of disorder, both intrinsic and extrinsic, especially for n-type materials. Intrinsic dynamic disorder stems from large thermal fluctuations both in intermolecular transfer integrals and (molecular) site energies in weakly interacting van der Waals solids and sources transient localization of the charge carriers. The molecular vibrations that drive transient localization typically operate at low-frequency (< a-few-hundred cm-1), which renders it difficult to assess them experimentally. Hitherto, this has prevented the identification of clear molecular design rules to control and reduce dynamic disorder. In addition, the disorder can also be extrinsic, being controlled by the gate insulator dielectric properties. Here we report on a comprehensive study of charge transport in two closely related n-type molecular organic semiconductors using a combination of temperature-dependent inelastic neutron scattering and photoelectron spectroscopy corroborated by electrical measurements, theory and simulations. We provide unambiguous evidence that ad hoc molecular design enables to free the electron charge carriers from both intrinsic and extrinsic disorder to ultimately reach band-like electron transport.

cond-mat.mtrl-sci↗

arXiv2019

Aesthetic Attributes Assessment of Images

Image aesthetic quality assessment has been a relatively hot topic during the last decade. Most recently, comments type assessment (aesthetic captions) has been proposed to describe the general aesthetic impression of an image using text. In this paper, we propose Aesthetic Attributes Assessment of Images, which means the aesthetic attributes captioning. This is a new formula of image aesthetic assessment, which predicts aesthetic attributes captions together with the aesthetic score of each attribute. We introduce a new dataset named \emph{DPC-Captions} which contains comments of up to 5 aesthetic attributes of one image through knowledge transfer from a full-annotated small-scale dataset. Then, we propose Aesthetic Multi-Attribute Network (AMAN), which is trained on a mixture of fully-annotated small-scale PCCD dataset and weakly-annotated large-scale DPC-Captions dataset. Our AMAN makes full use of transfer learning and attention model in a single framework. The experimental results on our DPC-Captions and PCCD dataset reveal that our method can predict captions of 5 aesthetic attributes together with numerical score assessment of each attribute. We use the evaluation criteria used in image captions to prove that our specially designed AMAN model outperforms traditional CNN-LSTM model and modern SCA-CNN model of image captions.

cs.CV↗

arXiv2019

Facial Makeup Transfer Combining Illumination Transfer

To meet the women appearance needs, we present a novel virtual experience approach of facial makeup transfer, developed into windows platform application software. The makeup effects could present on the user's input image in real time, with an only single reference image. The input image and reference image are divided into three layers by facial feature points landmarked: facial structure layer, facial color layer, and facial detail layer. Except for the above layers are processed by different algorithms to generate output image, we also add illumination transfer, so that the illumination effect of the reference image is automatically transferred to the input image. Our approach has the following three advantages: (1) Black or dark and white facial makeup could be effectively transferred by introducing illumination transfer; (2) Efficiently transfer facial makeup within seconds compared to those methods based on deep learning frameworks; (3) Reference images with the air-bangs could transfer makeup perfectly.

cs.CV↗

arXiv2019

A Survey and Experimental Analysis of Distributed Subgraph Matching

Recently there emerge many distributed algorithms that aim at solving subgraph matching at scale. Existing algorithm-level comparisons failed to provide a systematic view to the pros and cons of each algorithm mainly due to the intertwining of strategy and optimization. In this paper, we identify four strategies and three general-purpose optimizations from representative state-of-the-art works. We implement the four strategies with the optimizations based on the common Timely dataflow system for systematic strategy-level comparison. Our implementation covers all representation algorithms. We conduct extensive experiments for both unlabelled matching and labelled matching to analyze the performance of distributed subgraph matching under various settings, which is finally summarized as a practical guide.

cs.DB↗

arXiv2019

Flash: Efficient Dynamic Routing for Offchain Networks

Offchain networks emerge as a promising solution to address the scalability challenge of blockchain. Participants directly make payments through a network of payment channels without the overhead of committing onchain transactions. Routing is critical to the performance of offchain networks. Existing solutions use either static routing with poor performance or dynamic routing with high overhead to obtain the dynamic channel balance information. In this paper, we propose Flash, a new dynamic routing solution that leverages the unique characteristics of transactions in offchain networks to strike a better tradeoff between path optimality and probing overhead. By studying the traces of real offchain networks, we find that the payment sizes are heavy-tailed, and most payments are highly recurrent. Flash thus differentiates the treatment of elephant payments from mice payments. It uses a modified max-flow algorithm for elephant payments to find paths with sufficient capacity, and strategically routes the payment across paths to minimize the transaction fees. Mice payments are directly sent by looking up a routing table with a few precomputed paths to reduce probing overhead. Testbed experiments and data-driven simulations show that Flash improves the success volume of payments by up to 2.3x compared to the state-of-the-art routing algorithm.

cs.NI↗

arXiv2019

Robust quantitative single-exposure laser speckle imaging with true flow speckle contrast in the temporal and spatial domains

A systematic and robust laser speckle contrast imaging (LSCI) method and procedure is presented, covering the LSCI system calibration, static scattering removal, and measurement noise estimation and correction to obtain a true flow speckle contrast and the flow speed from single-exposure LSCI measurements. We advocate to use as the speckle contrast instead of the conventional contrast K as the former relates simply to the flow velocity and is with additive noise alone. We demonstrate the efficacy of the proposed true flow speckle contrast by imaging phantom flow at varying speeds, showing that (1) the proposed recipe greatly enhances the linear sensitivity of the flow index (inverse decorrelation time) and the linearity covers the full span of flow speeds from 0 mm/s to 40 mm/s; and (2) the true flow speed can be recovered regardless of the overlying static scattering layers and the type of speckle statistics (temporal or spatial). The fundamental difference between the apparent temporal and spatial speckle contrasts is further revealed. The flow index recovered in the spatial domain is much more susceptible to static scattering and exhibit a shorter linearity range than that obtained in the temporal domain. The proposed LSCI analysis framework paves the way to estimate the true flow speed in the wide array of laser speckle contrast imaging applications.

physics.app-ph↗

arXiv2019

Harmonia: Near-Linear Scalability for Replicated Storage with In-Network Conflict Detection

Distributed storage employs replication to mask failures and improve availability. However, these systems typically exhibit a hard tradeoff between consistency and performance. Ensuring consistency introduces coordination overhead, and as a result the system throughput does not scale with the number of replicas. We present Harmonia, a replicated storage architecture that exploits the capability of new-generation programmable switches to obviate this tradeoff by providing near-linear scalability without sacrificing consistency. To achieve this goal, Harmonia detects read-write conflicts in the network, which enables any replica to serve reads for objects with no pending writes. Harmonia implements this functionality at line rate, thus imposing no performance overhead. We have implemented a prototype of Harmonia on a cluster of commodity servers connected by a Barefoot Tofino switch, and have integrated it with Redis. We demonstrate the generality of our approach by supporting a variety of replication protocols, including primary-backup, chain replication, Viewstamped Replication, and NOPaxos. Experimental results show that Harmonia improves the throughput of these protocols by up to 10X for a replication factor of 10, providing near-linear scalability up to the limit of our testbed.

cs.DC↗

arXiv2019

Efficient Selection of Quasar Candidates Based on Optical and Infrared Photometric Data Using Machine Learning

We aim to select quasar candidates based on the two large survey databases, Pan-STARRS and AllWISE. Exploring the distribution of quasars and stars in the color spaces, we find that the combination of infrared and optical photometry is more conducive to select quasar candidates. Two new color criterions (yW1W2 and izW1W2) are constructed to distinguish quasars from stars efficiently. With izW1W2, 98.30% of star contamination is eliminated, while 99.50% of quasars are retained, at least to the magnitude limit of our training set of stars. Based on the optical and infrared color features, we put forward an efficient schema to select quasar candidates and high redshift quasar candidates, in which two machine learning algorithms (XGBoost and SVM) are implemented. The XGBoost and SVM classifiers have proven to be very effective with accuracy of 99.46% when 8Color as input pattern and default model parameters. Applying the two optimal classifiers to the unknown Pan-STARRS and AllWISE cross-matched data set, a total of 2,006,632 intersected sources are predicted to be quasar candidates given quasar probability larger than 0.5 (i.e. P_QSO>0.5). Among them, 1,201,211 have high probability (P_QSO>0.95). For these newly predicted quasar candidates, a regressor is constructed to estimate their redshifts. Finally 7,402 z>3.5 quasars are obtained. Given the magnitude limitation and site of the LAMOST telescope, part of these candidates will be used as the input catalogue of the LAMOST telescope for follow-up observation, and the rest may be observed by other telescopes.

astro-ph.IM↗

arXiv2019

Inverted Scanning Microwave Microscope for In Vitro Imaging and Characterization of Biological Cells

This paper presents for the first time an innovative instrument called an inverted scanning microwave microscope (iSMM), which is capable of noninvasive and label-free imaging and characterization of intracellular structures of a live cell on the nanometer scale. In particular, the iSMM is sensitive to not only surface structures, but also ectromagnetic properties up to one micrometer below the surface. Conveniently, the iSMM can be constructed through straightforward conversion of any scanning probe microscope, such as the atomic force microscope or the scanning tunneling microscope, with a simple metal probe to outperform traditional SMM in terms of ruggedness, and width, sensitivity and dynamic range. By contrast, the application of the traditional SMM to date has been limited to mainly surface physics and semiconductor technology, because the traditional SMM requires a fragile and expensive probe and is incompatible with saline solution or live biological cells.

physics.ins-det↗

arXiv2019

Neural Packet Classification

Packet classification is a fundamental problem in computer networking. This problem exposes a hard tradeoff between the computation and state complexity, which makes it particularly challenging. To navigate this tradeoff, existing solutions rely on complex hand-tuned heuristics, which are brittle and hard to optimize. In this paper, we propose a deep reinforcement learning (RL) approach to solve the packet classification problem. There are several characteristics that make this problem a good fit for Deep RL. First, many of the existing solutions are iteratively building a decision tree by splitting nodes in the tree. Second, the effects of these actions (e.g., splitting nodes) can only be evaluated once we are done with building the tree. These two characteristics are naturally captured by the ability of RL to take actions that have sparse and delayed rewards. Third, it is computationally efficient to generate data traces and evaluate decision trees, which alleviate the notoriously high sample complexity problem of Deep RL algorithms. Our solution, NeuroCuts, uses succinct representations to encode state and action space, and efficiently explore candidate decision trees to optimize for a global objective. It produces compact decision trees optimized for a specific set of rules and a given performance metric, such as classification time, memory footprint, or a combination of the two. Evaluation on ClassBench shows that NeuroCuts outperforms existing hand-crafted algorithms in classification time by 18% at the median, and reduces both time and memory footprint by up to 3x.

cs.NI↗