arXiv subjects

Xiaofeng Liu

Publications and source records attributed to Xiaofeng Liu.

173 records · Page 10Linked to original sources

arXiv2019

Attention Control with Metric Learning Alignment for Image Set-based Recognition

This paper considers the problem of image set-based face verification and identification. Unlike traditional single sample (an image or a video) setting, this situation assumes the availability of a set of heterogeneous collection of orderless images and videos. The samples can be taken at different check points, different identity documents $etc$. The importance of each image is usually considered either equal or based on a quality assessment of that image independent of other images and/or videos in that image set. How to model the relationship of orderless images within a set remains a challenge. We address this problem by formulating it as a Markov Decision Process (MDP) in a latent space. Specifically, we first propose a dependency-aware attention control (DAC) network, which uses actor-critic reinforcement learning for attention decision of each image to exploit the correlations among the unordered images. An off-policy experience replay is introduced to speed up the learning process. Moreover, the DAC is combined with a temporal model for videos using divide and conquer strategies. We also introduce a pose-guided representation (PGR) scheme that can further boost the performance at extreme poses. We propose a parameter-free PGR without the need for training as well as a novel metric learning-based PGR for pose alignment without the need for pose detection in testing stage. Extensive evaluations on IJB-A/B/C, YTF, Celebrity-1000 datasets demonstrate that our method outperforms many state-of-art approaches on the set-based as well as video-based face recognition databases.

cs.CV↗

arXiv2019

Permutation-invariant Feature Restructuring for Correlation-aware Image Set-based Recognition

We consider the problem of comparing the similarity of image sets with variable-quantity, quality and un-ordered heterogeneous images. We use feature restructuring to exploit the correlations of both inner$\&$inter-set images. Specifically, the residual self-attention can effectively restructure the features using the other features within a set to emphasize the discriminative images and eliminate the redundancy. Then, a sparse/collaborative learning-based dependency-guided representation scheme reconstructs the probe features conditional to the gallery features in order to adaptively align the two sets. This enables our framework to be compatible with both verification and open-set identification. We show that the parametric self-attention network and non-parametric dictionary learning can be trained end-to-end by a unified alternative optimization scheme, and that the full framework is permutation-invariant. In the numerical experiments we conducted, our method achieves top performance on competitive image set/video-based face recognition and person re-identification benchmarks.

cs.CV↗

arXiv2019

Dependency-aware Attention Control for Unconstrained Face Recognition with Image Sets

This paper targets the problem of image set-based face verification and identification. Unlike traditional single media (an image or video) setting, we encounter a set of heterogeneous contents containing orderless images and videos. The importance of each image is usually considered either equal or based on their independent quality assessment. How to model the relationship of orderless images within a set remains a challenge. We address this problem by formulating it as a Markov Decision Process (MDP) in the latent space. Specifically, we first present a dependency-aware attention control (DAC) network, which resorts to actor-critic reinforcement learning for sequential attention decision of each image embedding to fully exploit the rich correlation cues among the unordered images. Moreover, we introduce its sample-efficient variant with off-policy experience replay to speed up the learning process. The pose-guided representation scheme can further boost the performance at the extremes of the pose variation.

cs.CV↗

arXiv2018

Contextual-based Image Inpainting: Infer, Match, and Translate

We study the task of image inpainting, which is to fill in the missing region of an incomplete image with plausible contents. To this end, we propose a learning-based approach to generate visually coherent completion given a high-resolution image with missing components. In order to overcome the difficulty to directly learn the distribution of high-dimensional image data, we divide the task into inference and translation as two separate steps and model each step with a deep neural network. We also use simple heuristics to guide the propagation of local textures from the boundary to the hole. We show that, by using such techniques, inpainting reduces to the problem of learning two image-feature translation functions in much smaller space and hence easier to train. We evaluate our method on several public datasets and show that we generate results of better visual quality than previous state-of-the-art methods.

cs.CV↗

arXiv2018

Image Inpainting using Block-wise Procedural Training with Annealed Adversarial Counterpart

Recent advances in deep generative models have shown promising potential in image inpanting, which refers to the task of predicting missing pixel values of an incomplete image using the known context. However, existing methods can be slow or generate unsatisfying results with easily detectable flaws. In addition, there is often perceivable discontinuity near the holes and require further post-processing to blend the results. We present a new approach to address the difficulty of training a very deep generative model to synthesize high-quality photo-realistic inpainting. Our model uses conditional generative adversarial networks (conditional GANs) as the backbone, and we introduce a novel block-wise procedural training scheme to stabilize the training while we increase the network depth. We also propose a new strategy called adversarial loss annealing to reduce the artifacts. We further describe several losses specifically designed for inpainting and show their effectiveness. Extensive experiments and user-study show that our approach outperforms existing methods in several tasks such as inpainting, face completion and image harmonization. Finally, we show our framework can be easily used as a tool for interactive guided inpainting, demonstrating its practical value to solve common real-world challenges.

cs.CV↗

arXiv2017

Sonication-assisted synthesis of semiconducting atomic crystals of germanium monoselenide: An isostructural and isoelectronic analog of black phosphorous

Monochalcogenides of germanium (or tin) are considered as stable isoelectronic and isostructural analogue of black phosphorous. Their two-dimensional (2D) forms have been just predicted to shown strong thickness-dependent physical properties, and even indirect to direct band gap crossover at the monolayer limit. Here, we demonstrate the synthesis of atomically thin GeSe by direct sonication-assisted exfoliation of bulk microcrystalline powders in solvents. The exfoliated few-layer GeSe sheets characterize high crystallinity with lateral sizes over 100 nm and, importantly, strong resistance against oxidation and degradation in ambient conditions. Density functional theory calculation combined with optical characterizations confirm the layer-number dependent optical bandgap, which, for few-layer sheets, is optimal for solar light harvesting, and promising for relevant applications, such as optoelectronics and photonics.

cond-mat.mtrl-sci↗

arXiv2017

Exploiting ITO colloidal nanocrystals for ultrafast pulse generation

Dynamical materials that capable of responding to optical stimuli have always been pursued for designing novel photonic devices and functionalities, of which the response speed and amplitude as well as integration adaptability and energy effectiveness are especially critical. Here we show ultrafast pulse generation by exploiting the ultrafast and sensitive nonlinear dynamical processes in tunably solution-processed colloidal epsilon-near-zero (ENZ) transparent conducting oxide (TCO) nanocrystals (NCs), of which the potential respond response speed is >2 THz and modulation depth is ~23% pumped at ~0.7 mJ/cm2, benefiting from the highly confined geometry in addition to the ENZ enhancement effect. These ENZ NCs may offer a scalable and printable material solution for dynamic photonic and optoelectronic devices.

cond-mat.mtrl-sci↗

arXiv2015

Fast Online Clustering with Randomized Skeleton Sets

We present a new fast online clustering algorithm that reliably recovers arbitrary-shaped data clusters in high throughout data streams. Unlike the existing state-of-the-art online clustering methods based on k-means or k-medoid, it does not make any restrictive generative assumptions. In addition, in contrast to existing nonparametric clustering techniques such as DBScan or DenStream, it gives provable theoretical guarantees. To achieve fast clustering, we propose to represent each cluster by a skeleton set which is updated continuously as new data is seen. A skeleton set consists of weighted samples from the data where weights encode local densities. The size of each skeleton set is adapted according to the cluster geometry. The proposed technique automatically detects the number of clusters and is robust to outliers. The algorithm works for the infinite data stream where more than one pass over the data is not feasible. We provide theoretical guarantees on the quality of the clustering and also demonstrate its advantage over the existing state-of-the-art on several datasets.

cs.AI↗

arXiv2015

Increment entropy as a measure of complexity for time series

Entropy has been a common index to quantify the complexity of time series in a variety of fields. Here, we introduce increment entropy to measure the complexity of time series in which each increment is mapped into a word of two letters, one letter corresponding to direction and the other corresponding to magnitude. The Shannon entropy of the words is termed as increment entropy (IncrEn). Simulations on synthetic data and tests on epileptic EEG signals have demonstrated its ability of detecting the abrupt change, regardless of energetic (e.g. spikes or bursts) or structural changes. The computation of IncrEn does not make any assumption on time series and it can be applicable to arbitrary real-world data.

physics.data-an↗

arXiv2014

Tortuosity Entropy: a measure of spatial complexity of behavioral changes in animal movement data

The goal of animal movement analysis is to understand how organisms explore and exploit the complex and varying environment. Animals usually exhibit varied and complicated movements, from apparently deterministic behaviors to highly random ones. This is critical for assessing movement efficiency and strategies that are used to quantify and analyze movement trajectories. Here we introduce a tortuosity entropy (TorEn) based on comparison of parameters, e.g. heading, bearing, speed, of consecutive points in movement trajectory, which is a simple measure for quantifying the behavioral change in animal movement data in a fine scale. In our approach, the differences between pairwise successive track points are transformed inot symbolic sequences, then we map these symbols into a group of pattern vectors and calculate the information entropy of pattern vector. Tortuosity entropy can be easily applied to arbitrary real-world data-deterministic or stochastic, stationary or non-stationary. We test the algorithm on both simulated trajectories and real trajectories and show that both mixed segments in synthetic data and different phases in real movement data are identified accurately. The results show that the algorithm is applicable to various situations, indicating that our approach is a promising tool to reveal the behavioral pattern in movement data.

q-bio.QM↗

arXiv2011

Interactions of Shear Layer Vortices with the Trailing Corner in an Open Cavity Flow

This fluid dynamics video provides sample experimental results focusing on the interactions of shear layer vortices with the trailing corner in a 2D open cavity shear layer. These interactions were investigated experimentally in a water tunnel at a Reynolds number of 40,000. Time-resolved particle image velocimetry (PIV) with an image sampling rate of 4500 frames per second was used to simultaneously measure the instantaneous velocity, material acceleration and pressure distributions. The latter was calculated by integrating the spatial distribution of in-plane components of the material acceleration. A large database of instantaneous realizations visualized the dynamic changes to the shear layer vortices, such as deformation and breakup as they impinged and climbed over the cavity trailing corner. These interactions cause time-dependent formation of a pressure maximum as the flow impinges on the forward facing surface of the trailing corner, and a minimum above the corner, where large local pressure gradients dominate the generation of fresh vorticity. Data analysis shows that interactions with the corner involve two time scales. Intermediate Strouhal numbers (0.6-3.2) are associated with interaction of the shear layer vortices with the corner, while vertical flapping of the shear layer occurs at low Strouhal numbers (0.1-0.6).

physics.flu-dyn↗