arXiv subjects

Jing Zhang

Publications and source records attributed to Jing Zhang.

At least 757 records · Page 42Linked to original sources

arXiv2021

Privacy-Preserving Portrait Matting

Recently, there has been an increasing concern about the privacy issue raised by using personally identifiable information in machine learning. However, previous portrait matting methods were all based on identifiable portrait images. To fill the gap, we present P3M-10k in this paper, which is the first large-scale anonymized benchmark for Privacy-Preserving Portrait Matting. P3M-10k consists of 10,000 high-resolution face-blurred portrait images along with high-quality alpha mattes. We systematically evaluate both trimap-free and trimap-based matting methods on P3M-10k and find that existing matting methods show different generalization capabilities when following the Privacy-Preserving Training (PPT) setting, i.e., training on face-blurred images and testing on arbitrary images. To devise a better trimap-free portrait matting model, we propose P3M-Net, which leverages the power of a unified framework for both semantic perception and detail matting, and specifically emphasizes the interaction between them and the encoder to facilitate the matting process. Extensive experiments on P3M-10k demonstrate that P3M-Net outperforms the state-of-the-art methods in terms of both objective metrics and subjective visual quality. Besides, it shows good generalization capacity under the PPT setting, confirming the value of P3M-10k for facilitating future research and enabling potential real-world applications. The source code and dataset are available at https://github.com/JizhiziLi/P3M

cs.CV↗

arXiv2021

Active Learning to Classify Macromolecular Structures in situ for Less Supervision in Cryo-Electron Tomography

Motivation: Cryo-Electron Tomography (cryo-ET) is a 3D bioimaging tool that visualizes the structural and spatial organization of macromolecules at a near-native state in single cells, which has broad applications in life science. However, the systematic structural recognition and recovery of macromolecules captured by cryo-ET are difficult due to high structural complexity and imaging limits. Deep learning based subtomogram classification have played critical roles for such tasks. As supervised approaches, however, their performance relies on sufficient and laborious annotation on a large training dataset. Results: To alleviate this major labeling burden, we proposed a Hybrid Active Learning (HAL) framework for querying subtomograms for labelling from a large unlabeled subtomogram pool. Firstly, HAL adopts uncertainty sampling to select the subtomograms that have the most uncertain predictions. Moreover, to mitigate the sampling bias caused by such strategy, a discriminator is introduced to judge if a certain subtomogram is labeled or unlabeled and subsequently the model queries the subtomogram that have higher probabilities to be unlabeled. Additionally, HAL introduces a subset sampling strategy to improve the diversity of the query set, so that the information overlap is decreased between the queried batches and the algorithmic efficiency is improved. Our experiments on subtomogram classification tasks using both simulated and real data demonstrate that we can achieve comparable testing performance (on average only 3% accuracy drop) by using less than 30% of the labeled subtomograms, which shows a very promising result for subtomogram classification task with limited labeling resources.

q-bio.QM↗

arXiv2021

DSP: Dual Soft-Paste for Unsupervised Domain Adaptive Semantic Segmentation

Unsupervised domain adaptation (UDA) for semantic segmentation aims to adapt a segmentation model trained on the labeled source domain to the unlabeled target domain. Existing methods try to learn domain invariant features while suffering from large domain gaps that make it difficult to correctly align discrepant features, especially in the initial training phase. To address this issue, we propose a novel Dual Soft-Paste (DSP) method in this paper. Specifically, DSP selects some classes from a source domain image using a long-tail class first sampling strategy and softly pastes the corresponding image patch on both the source and target training images with a fusion weight. Technically, we adopt the mean teacher framework for domain adaptation, where the pasted source and target images go through the student network while the original target image goes through the teacher network. Output-level alignment is carried out by aligning the probability maps of the target fused image from both networks using a weighted cross-entropy loss. In addition, feature-level alignment is carried out by aligning the feature maps of the source and target images from student network using a weighted maximum mean discrepancy loss. DSP facilitates the model learning domain-invariant features from the intermediate domains, leading to faster convergence and better performance. Experiments on two challenging benchmarks demonstrate the superiority of DSP over state-of-the-art methods. Code is available at \url{https://github.com/GaoLii/DSP}.

cs.CV↗

arXiv2021

Spectral Integrals of Bernoulli Generalized Functionals

Let $\mathcal{S}\subset \mathcal{L}^2 \subset \mathcal{S}^*$ be the Gel'fand triple over the Bernoulli space, where elements of $\mathcal{S}^*$ are called Bernoulli generalized functionals. In this paper, we define integrals of Bernoulli generalized functionals with respect to a spectral measure (projection operator-valued measure) in the framework of $\mathcal{S}\subset \mathcal{L}^2 \subset \mathcal{S}^*$, and examine their fundamental properties. New notions are introduced, several results are obtained and examples are also shown.

math.FA↗

arXiv2021

Deep Automatic Natural Image Matting

Automatic image matting (AIM) refers to estimating the soft foreground from an arbitrary natural image without any auxiliary input like trimap, which is useful for image editing. Prior methods try to learn semantic features to aid the matting process while being limited to images with salient opaque foregrounds such as humans and animals. In this paper, we investigate the difficulties when extending them to natural images with salient transparent/meticulous foregrounds or non-salient foregrounds. To address the problem, a novel end-to-end matting network is proposed, which can predict a generalized trimap for any image of the above types as a unified semantic representation. Simultaneously, the learned semantic features guide the matting network to focus on the transition areas via an attention mechanism. We also construct a test set AIM-500 that contains 500 diverse natural images covering all types along with manually labeled alpha mattes, making it feasible to benchmark the generalization ability of AIM models. Results of the experiments demonstrate that our network trained on available composite matting datasets outperforms existing methods both objectively and subjectively. The source code and dataset are available at https://github.com/JizhiziLi/AIM.

cs.CV↗

arXiv2021

Imaging the dipole scattering of an optically levitated dielectric nanoparticle

We experimentally observe the dipole scattering of a nanoparticle using a high numerical aperture (NA) imaging system. The optically levitated nanoparticle provides an environment free of particle-substrate interaction. We illuminate the silica nanoparticle in vacuum with a 532 nm laser beam orthogonally to the propagation direction of the 1064 nm trapping laser beam strongly focused by the same high NA objective used to collect the scattering, which results in a dark background and high signal-noise ratio. The dipole orientations of the nanoparticle induced by the linear polarization of the incident laser are studied by measuring the scattering light distribution in the image and the Fourier space (k-space) as we rotate the illuminating light polarization. The polarization vortex (vector beam) is observed for the special case, when the dipole orientation of the nanoparticle is aligned along the optical axis of the microscope objective. Our work offers an important platform for studying the scattering anisotropy with Kerker conditions.

physics.optics↗

arXiv2021

One-Shot Affordance Detection

Affordance detection refers to identifying the potential action possibilities of objects in an image, which is an important ability for robot perception and manipulation. To empower robots with this ability in unseen scenarios, we consider the challenging one-shot affordance detection problem in this paper, i.e., given a support image that depicts the action purpose, all objects in a scene with the common affordance should be detected. To this end, we devise a One-Shot Affordance Detection (OS-AD) network that firstly estimates the purpose and then transfers it to help detect the common affordance from all candidate images. Through collaboration learning, OS-AD can capture the common characteristics between objects having the same underlying affordance and learn a good adaptation capability for perceiving unseen affordances. Besides, we build a Purpose-driven Affordance Dataset (PAD) by collecting and labeling 4k images from 31 affordance and 72 object categories. Experimental results demonstrate the superiority of our model over previous representative ones in terms of both objective metrics and visual quality. The benchmark suite is at ProjectPage.

cs.CV↗

arXiv2021

Invertible Attention

Attention has been proved to be an efficient mechanism to capture long-range dependencies. However, so far it has not been deployed in invertible networks. This is due to the fact that in order to make a network invertible, every component within the network needs to be a bijective transformation, but a normal attention block is not. In this paper, we propose invertible attention that can be plugged into existing invertible models. We mathematically and experimentally prove that the invertibility of an attention model can be achieved by carefully constraining its Lipschitz constant. We validate the invertibility of our invertible attention on image reconstruction task with 3 popular datasets: CIFAR-10, SVHN, and CelebA. We also show that our invertible attention achieves similar performance in comparison with normal non-invertible attention on dense prediction tasks. The code is available at https://github.com/Schwartz-Zha/InvertibleAttention

cs.CV↗

arXiv2021

Few-Shot Domain Expansion for Face Anti-Spoofing

Face anti-spoofing (FAS) is an indispensable and widely used module in face recognition systems. Although high accuracy has been achieved, a FAS system will never be perfect due to the non-stationary applied environments and the potential emergence of new types of presentation attacks in real-world applications. In practice, given a handful of labeled samples from a new deployment scenario (target domain) and abundant labeled face images in the existing source domain, the FAS system is expected to perform well in the new scenario without sacrificing the performance on the original domain. To this end, we identify and address a more practical problem: Few-Shot Domain Expansion for Face Anti-Spoofing (FSDE-FAS). This problem is challenging since with insufficient target domain training samples, the model may suffer from both overfitting to the target domain and catastrophic forgetting of the source domain. To address the problem, this paper proposes a Style transfer-based Augmentation for Semantic Alignment (SASA) framework. We propose to augment the target data by generating auxiliary samples based on photorealistic style transfer. With the assistant of the augmented data, we further propose a carefully designed mechanism to align different domains from both instance-level and distribution-level, and then stabilize the performance on the source domain with a less-forgetting constraint. Two benchmarks are proposed to simulate the FSDE-FAS scenarios, and the experimental results show that the proposed SASA method outperforms state-of-the-art methods.

cs.CV↗

arXiv2021

Confidence-Aware Learning for Camouflaged Object Detection

Confidence-aware learning is proven as an effective solution to prevent networks becoming overconfident. We present a confidence-aware camouflaged object detection framework using dynamic supervision to produce both accurate camouflage map and meaningful "confidence" representing model awareness about the current prediction. A camouflaged object detection network is designed to produce our camouflage prediction. Then, we concatenate it with the input image and feed it to the confidence estimation network to produce an one channel confidence map.We generate dynamic supervision for the confidence estimation network, representing the agreement of camouflage prediction with the ground truth camouflage map. With the produced confidence map, we introduce confidence-aware learning with the confidence map as guidance to pay more attention to the hard/low-confidence pixels in the loss function. We claim that, once trained, our confidence estimation network can evaluate pixel-wise accuracy of the prediction without relying on the ground truth camouflage map. Extensive results on four camouflaged object detection testing datasets illustrate the superior performance of the proposed model in explaining the camouflage prediction.

cs.CV↗

arXiv2021

SIR: Self-supervised Image Rectification via Seeing the Same Scene from Multiple Different Lenses

Deep learning has demonstrated its power in image rectification by leveraging the representation capacity of deep neural networks via supervised training based on a large-scale synthetic dataset. However, the model may overfit the synthetic images and generalize not well on real-world fisheye images due to the limited universality of a specific distortion model and the lack of explicitly modeling the distortion and rectification process. In this paper, we propose a novel self-supervised image rectification (SIR) method based on an important insight that the rectified results of distorted images of a same scene from different lens should be the same. Specifically, we devise a new network architecture with a shared encoder and several prediction heads, each of which predicts the distortion parameter of a specific distortion model. We further leverage a differentiable warping module to generate the rectified images and re-distorted images from the distortion parameters and exploit the intra- and inter-model consistency between them during training, thereby leading to a self-supervised learning scheme without the need for ground-truth distortion parameters or normal images. Experiments on synthetic dataset and real-world fisheye images demonstrate that our method achieves comparable or even better performance than the supervised baseline method and representative state-of-the-art methods. Self-supervised learning also improves the universality of distortion models while keeping their self-consistency.

cs.CV↗

arXiv2021

Quasi-solid-state sodium-ion hybrid capacitors enabled by UiO-66@PVDF-HFP multifunctional separators: selective charge transfer and high safety

The practical application of sodium-ion hybrid capacitors is limited by their low energy densities resulted from the kinetics mismatch between cathodes and anodes, and the fire safety related to the flammable electrolyte-separator system. Hence, we report a rational design of metal-organic frameworks (MOFs, UiO-66) modified PVDF-HFP separator. High tensile strength and dimensional thermal stability of the separator reduce the risk of electrode short circuit caused by the separator deformation. MCC test demonstrates a reduction of 75% in peak heat release rate (pHRR), indicating an enhanced fire-resistant property of the separator. This is due to the transformation of UiO-66 into ZrO2 accompanied by the consumption of oxygen and the formation of the barrier char that suppresses further heat release. Quasi-solid-state electrolyte prepared based on this separator presents an enhanced ionic conductivity of 2.44 mS*cm-1 and Na-ion transference number of 0.55, which are related to the high porosity ( >70%) and electrolyte uptake (~ 320%) of the separator. Moreover, the open metal sites of UiO-66 can capture PF6- and consequently liberate the Na+ for faster migration, thus reducing the kinetics mismatch between cathodes and anodes. Such multifunctional separator enables the quasi-solid-state Na-ion hybrid capacitor to achieve high energy density (182 Wh*kg-1 @31 W*kg-1) and power density (5280 W*kg-1 @22 Wh*kg-1), as well as excellent cyclic stability (10000 cycles @1000 mA*g-1). Keywords: Quasi-solid-state; PVDF-HFP; Metal-organic frameworks; Dimensional thermal stability; Fire safety; Selective charge transfer

physics.app-ph↗

arXiv2021

Towards High Performance Human Keypoint Detection

Human keypoint detection from a single image is very challenging due to occlusion, blur, illumination and scale variance. In this paper, we address this problem from three aspects by devising an efficient network structure, proposing three effective training strategies, and exploiting four useful postprocessing techniques. First, we find that context information plays an important role in reasoning human body configuration and invisible keypoints. Inspired by this, we propose a cascaded context mixer (CCM), which efficiently integrates spatial and channel context information and progressively refines them. Then, to maximize CCM's representation capability, we develop a hard-negative person detection mining strategy and a joint-training strategy by exploiting abundant unlabeled data. It enables CCM to learn discriminative features from massive diverse poses. Third, we present several sub-pixel refinement techniques for postprocessing keypoint predictions to improve detection accuracy. Extensive experiments on the MS COCO keypoint detection benchmark demonstrate the superiority of the proposed method over representative state-of-the-art (SOTA) methods. Our single model achieves comparable performance with the winner of the 2018 COCO Keypoint Detection Challenge. The final ensemble model sets a new SOTA on this benchmark.

cs.CV↗

arXiv2021

Stochastic Differential Games with Random Coefficients and Stochastic Hamilton-Jacobi-Bellman-Isaacs Equations

In this paper, we study a class of zero-sum two-player stochastic differential games with the controlled stochastic differential equations and the payoff/cost functionals of recursive type. As opposed to the pioneering work by Fleming and Souganidis [Indiana Univ. Math. J., 38 (1989), pp.~293--314] and the seminal work by Buckdahn and Li [SIAM J. Control Optim., 47 (2008), pp.~444--475], the involved coefficients may be random, going beyond the Markovian framework and leading to the random upper and lower value functions. We first prove the dynamic programming principle for the game, and then under the standard Lipschitz continuity assumptions on the coefficients, the upper and lower value functions are shown to be the viscosity solutions of the upper and the lower fully nonlinear stochastic Hamilton-Jacobi-Bellman-Isaacs (HJBI) equations, respectively. A stability property of viscosity solutions is also proved. Under certain additional regularity assumptions on the diffusion coefficient, the uniqueness of the viscosity solution is addressed as well.

math.PR↗

arXiv2021

Progressive One-shot Human Parsing

Prior human parsing models are limited to parsing humans into classes pre-defined in the training data, which is not flexible to generalize to unseen classes, e.g., new clothing in fashion analysis. In this paper, we propose a new problem named one-shot human parsing (OSHP) that requires to parse human into an open set of reference classes defined by any single reference example. During training, only base classes defined in the training set are exposed, which can overlap with part of reference classes. In this paper, we devise a novel Progressive One-shot Parsing network (POPNet) to address two critical challenges , i.e., testing bias and small sizes. POPNet consists of two collaborative metric learning modules named Attention Guidance Module and Nearest Centroid Module, which can learn representative prototypes for base classes and quickly transfer the ability to unseen classes during testing, thereby reducing testing bias. Moreover, POPNet adopts a progressive human parsing framework that can incorporate the learned knowledge of parent classes at the coarse granularity to help recognize the descendant classes at the fine granularity, thereby handling the small sizes issue. Experiments on the ATR-OS benchmark tailored for OSHP demonstrate POPNet outperforms other representative one-shot segmentation models by large margins and establishes a strong baseline. Source code can be found at https://github.com/Charleshhy/One-shot-Human-Parsing.

cs.CV↗

arXiv2021

Recyclable flame-retardant epoxy composites based on disulfide bonds. Flammability and recyclability

A series of recyclable epoxy resin (REP) composites are successfully prepared in this study, aiming to improve flame retardancy while keep good recyclability and mechanical properties. The commercial 9,10-Dihydro-9-oxa-10-phosphaphenanthrene 10-Oxide (DOPO) with high reactive P-H bond is used as highly-efficient flame retardant in this work. Compared with pure epoxy resin with a limited oxygen index of 21.7 % and no rating in vertical burning test, recyclable epoxy resin only with 3 wt% DOPO acquires a limited oxygen index value of 27.8% and V-0 rating. In addition, REP composites containing 4 wt% DOPO displays quick self-extinguishing behavior, which shows the 43.5% drop of the peak heat release rate. The main reason might be that traditional phosphorous-containing flame retardant existing in REP composites can play a positive role in both condensed phase and gas phase when burning. On the other hand, both scanning electron microscopy and Raman results also further affirm the improved flame retardancy. Moreover, flame-retardant REP composites can also maintain good transparency and mechanical properties. As well as the favorable recyclability originated from the special exchangeable disulfide crosslinks. High-Performance REP composites possessing expected flame retardancy, recyclability and good mechanical properties pave the way to broaden applications of thermoset epoxy materials.

physics.app-ph↗

arXiv2021

Strongest transformations

We continue our study of maps transforming high-dimensional complicated objects into squares of stationary sets. Previously, we proved that many such transformations exist in ZFC, and here we address the consistency of the strongest conceivable transformations. Along the way, we obtain new results on Shelah's coloring principle $Pr_1$. For $κ$ inaccessible, we prove the consistency of $Pr_1(κ,κ,κ,κ)$. For successors of regulars, we obtain a full lifting of Galvin's 1980 theorem. In contrast, the full lifting of Galvin's theorem to successors of singulars is shown to be inconsistent.

math.LO↗

arXiv2021

6 GHz hyperfast rotation of an optically levitated nanoparticle in vacuum

We report an experimental observation of a record-breaking ultra-high rotation frequency about 6 GHz in an optically levitated nanoparticle system. We optically trap a nanoparticle in the gravity direction with a high numerical aperture (NA) objective lens, which shows significant advantages in compensating the influences of the scattering force and the photophoretic force on the trap, especially at intermediate pressure (about 100 Pa). This allows us to trap a nanoparticle from atmospheric to low pressure ($10^{-3}$ Pa) without using feedback cooling. We measure a highest rotation frequency about 4.3 GHz of the trapped nanoparticle without feedback cooling and a 6 GHz rotation with feedback cooling, which is the fastest mechanical rotation ever reported to date. Our work provides useful guides for efficiently observing hyperfast rotation in the optical levitation system, and may find various applications such as in ultrasensitive torque detection, probing vacuum friction, and testing unconventional decoherence theories.

physics.optics↗