arXiv subjects

Fang Zhao

Publications and source records attributed to Fang Zhao.

71 records · Page 4Linked to original sources

arXiv2021

Learning Anchored Unsigned Distance Functions with Gradient Direction Alignment for Single-view Garment Reconstruction

While single-view 3D reconstruction has made significant progress benefiting from deep shape representations in recent years, garment reconstruction is still not solved well due to open surfaces, diverse topologies and complex geometric details. In this paper, we propose a novel learnable Anchored Unsigned Distance Function (AnchorUDF) representation for 3D garment reconstruction from a single image. AnchorUDF represents 3D shapes by predicting unsigned distance fields (UDFs) to enable open garment surface modeling at arbitrary resolution. To capture diverse garment topologies, AnchorUDF not only computes pixel-aligned local image features of query points, but also leverages a set of anchor points located around the surface to enrich 3D position features for query points, which provides stronger 3D space context for the distance function. Furthermore, in order to obtain more accurate point projection direction at inference, we explicitly align the spatial gradient direction of AnchorUDF with the ground-truth direction to the surface during training. Extensive experiments on two public 3D garment datasets, i.e., MGN and Deep Fashion3D, demonstrate that AnchorUDF achieves the state-of-the-art performance on single-view garment reconstruction.

cs.CV↗

arXiv2021

DomainMix: Learning Generalizable Person Re-Identification Without Human Annotations

Existing person re-identification models often have low generalizability, which is mostly due to limited availability of large-scale labeled data in training. However, labeling large-scale training data is very expensive and time-consuming, while large-scale synthetic dataset shows promising value in learning generalizable person re-identification models. Therefore, in this paper a novel and practical person re-identification task is proposed,i.e. how to use labeled synthetic dataset and unlabeled real-world dataset to train a universal model. In this way, human annotations are no longer required, and it is scalable to large and diverse real-world datasets. To address the task, we introduce a framework with high generalizability, namely DomainMix. Specifically, the proposed method firstly clusters the unlabeled real-world images and selects the reliable clusters. During training, to address the large domain gap between two domains, a domain-invariant feature learning method is proposed, which introduces a new loss,i.e. domain balance loss, to conduct an adversarial learning between domain-invariant feature learning and domain discrimination, and meanwhile learns a discriminative feature for person re-identification. This way, the domain gap between synthetic and real-world data is much reduced, and the learned feature is generalizable thanks to the large-scale and diverse training data. Experimental results show that the proposed annotation-free method is more or less comparable to the counterpart trained with full human annotations, which is quite promising. In addition, it achieves the current state of the art on several person re-identification datasets under direct cross-dataset evaluation.

cs.CV↗

arXiv2021

Lvio-Fusion: A Self-adaptive Multi-sensor Fusion SLAM Framework Using Actor-critic Method

State estimation with sensors is essential for mobile robots. Due to different performance of sensors in different environments, how to fuse measurements of various sensors is a problem. In this paper, we propose a tightly coupled multi-sensor fusion framework, Lvio-Fusion, which fuses stereo camera, Lidar, IMU, and GPS based on the graph optimization. Especially for urban traffic scenes, we introduce a segmented global pose graph optimization with GPS and loop-closure, which can eliminate accumulated drifts. Additionally, we creatively use a actor-critic method in reinforcement learning to adaptively adjust sensors' weight. After training, actor-critic agent can provide the system better and dynamic sensors' weight. We evaluate the performance of our system on public datasets and compare it with other state-of-the-art methods, which shows that the proposed method achieves high estimation accuracy and robustness to various environments. And our implementations are open source and highly scalable.

cs.RO↗

arXiv2021

Beyond Monocular Deraining: Parallel Stereo Deraining Network Via Semantic Prior

Rain is a common natural phenomenon. Taking images in the rain however often results in degraded quality of images, thus compromises the performance of many computer vision systems. Most existing de-rain algorithms use only one single input image and aim to recover a clean image. Few work has exploited stereo images. Moreover, even for single image based monocular deraining, many current methods fail to complete the task satisfactorily because they mostly rely on per pixel loss functions and ignore semantic information. In this paper, we present a Paired Rain Removal Network (PRRNet), which exploits both stereo images and semantic information. Specifically, we develop a Semantic-Aware Deraining Module (SADM) which solves both tasks of semantic segmentation and deraining of scenes, and a Semantic-Fusion Network (SFNet) and a View-Fusion Network (VFNet) which fuse semantic information and multi-view information respectively. In addition, we also introduce an Enhanced Paired Rain Removal Network (EPRRNet) which exploits semantic prior to remove rain streaks from stereo images. We first use a coarse deraining network to reduce the rain streaks on the input images, and then adopt a pre-trained semantic segmentation network to extract semantic features from the coarse derained image. Finally, a parallel stereo deraining network fuses semantic and multi-view information to restore finer results. We also propose new stereo based rainy datasets for benchmarking. Experiments on both monocular and the newly proposed stereo rainy datasets demonstrate that the proposed method achieves the state-of-the-art performance.

cs.CV↗

arXiv2020

Quantum battery of interacting spins with environmental noise

A quantum battery is a temporary energy-storage system. We constructed the quantum battery model of an N-spin chain with nearest-neighbor hopping interaction and investigated the charging process of the quantum battery. We obtained the maximum energy in the quantum battery charged by a coherent cavity driving field or a thermal heat bath. We confirmed that for a finite-length spin chain, thermal charging results in a nonzero ergotropy, contradicting a previous result: that an incoherent heat source cannot charge a single-spin quantum battery. The nearest-neighbor hopping interaction induces energy band splitting, which enhances the energy storage and the ergotropy of the quantum battery. We found a critical point in the energy and ergotropy resulting from the ground-state quantum phase transition, after which the energy significantly enhance. Finally, we also found that disorder increased the energy of the quantum battery.

quant-ph↗

arXiv2019

Multi-Prototype Networks for Unconstrained Set-based Face Recognition

In this paper, we study the challenging unconstrained set-based face recognition problem where each subject face is instantiated by a set of media (images and videos) instead of a single image. Naively aggregating information from all the media within a set would suffer from the large intra-set variance caused by heterogeneous factors (e.g., varying media modalities, poses and illuminations) and fail to learn discriminative face representations. A novel Multi-Prototype Network (MPNet) model is thus proposed to learn multiple prototype face representations adaptively from the media sets. Each learned prototype is representative for the subject face under certain condition in terms of pose, illumination and media modality. Instead of handcrafting the set partition for prototype learning, MPNet introduces a Dense SubGraph (DSG) learning sub-net that implicitly untangles inconsistent media and learns a number of representative prototypes. Qualitative and quantitative experiments clearly demonstrate superiority of the proposed model over state-of-the-arts.

cs.CV↗

arXiv2018

Look Across Elapse: Disentangled Representation Learning and Photorealistic Cross-Age Face Synthesis for Age-Invariant Face Recognition

Despite the remarkable progress in face recognition related technologies, reliably recognizing faces across ages still remains a big challenge. The appearance of a human face changes substantially over time, resulting in significant intra-class variations. As opposed to current techniques for age-invariant face recognition, which either directly extract age-invariant features for recognition, or first synthesize a face that matches target age before feature extraction, we argue that it is more desirable to perform both tasks jointly so that they can leverage each other. To this end, we propose a deep Age-Invariant Model (AIM) for face recognition in the wild with three distinct novelties. First, AIM presents a novel unified deep architecture jointly performing cross-age face synthesis and recognition in a mutual boosting way. Second, AIM achieves continuous face rejuvenation/aging with remarkable photorealistic and identity-preserving properties, avoiding the requirement of paired data and the true age of testing samples. Third, we develop effective and novel training strategies for end-to-end learning the whole deep architecture, which generates powerful age-invariant face representations explicitly disentangled from the age variation. Moreover, we propose a new large-scale Cross-Age Face Recognition (CAFR) benchmark dataset to facilitate existing efforts and push the frontiers of age-invariant face recognition research. Extensive experiments on both our CAFR and several other cross-age datasets (MORPH, CACD and FG-NET) demonstrate the superiority of the proposed AIM model over the state-of-the-arts. Benchmarking our model on one of the most popular unconstrained face recognition datasets IJB-C additionally verifies the promising generalizability of AIM in recognizing faces in the wild.

cs.CV↗

arXiv2018

Discriminative Learning-based Smartphone Indoor Localization

Due to the growing area of ubiquitous mobile applications, indoor localization of smartphones has become an interesting research topic. Most of the current indoor localization systems rely on intensive site survey to achieve high accuracy. In this work, we propose an efficient smartphones indoor localization system that is able to reduce the site survey effort while still achieving high localization accuracy. Our system is built by fusing a variety of signals, such as Wi-Fi received signal strength indicator, magnetic field and floor plan information in an enhanced particle filter. To achieve high and stable performance, we first apply discriminative learning models to integrate Wi-Fi and magnetic field readings to achieve room level landmark detection. Further, we integrate landmark detection, range-based localization models, with a graph-based discretized system state representation. Because our approach requires only discriminative learning-based room level landmark detections, the time spent in the learning phase is significantly reduced compared to traditional Wi-Fi fingerprinting or landmark-based approaches. We conduct experimental studies to evaluate our system in an office-like indoor environment. Experiment results show that our system can significantly reduce the learning efforts, and the localization method can achieve performance with an average localization error of 1.55 meters.

cs.NI↗

arXiv2017

Integrated Face Analytics Networks through Cross-Dataset Hybrid Training

Face analytics benefits many multimedia applications. It consists of a number of tasks, such as facial emotion recognition and face parsing, and most existing approaches generally treat these tasks independently, which limits their deployment in real scenarios. In this paper we propose an integrated Face Analytics Network (iFAN), which is able to perform multiple tasks jointly for face analytics with a novel carefully designed network architecture to fully facilitate the informative interaction among different tasks. The proposed integrated network explicitly models the interactions between tasks so that the correlations between tasks can be fully exploited for performance boost. In addition, to solve the bottleneck of the absence of datasets with comprehensive training data for various tasks, we propose a novel cross-dataset hybrid training strategy. It allows "plug-in and play" of multiple datasets annotated for different tasks without the requirement of a fully labeled common dataset for all the tasks. We experimentally show that the proposed iFAN achieves state-of-the-art performance on multiple face analytics tasks using a single integrated model. Specifically, iFAN achieves an overall F-score of 91.15% on the Helen dataset for face parsing, a normalized mean error of 5.81% on the MTFL dataset for facial landmark localization and an accuracy of 45.73% on the BNU dataset for emotion recognition with a single model.

cs.CV↗

arXiv2016

Robust LSTM-Autoencoders for Face De-Occlusion in the Wild

Face recognition techniques have been developed significantly in recent years. However, recognizing faces with partial occlusion is still challenging for existing face recognizers which is heavily desired in real-world applications concerning surveillance and security. Although much research effort has been devoted to developing face de-occlusion methods, most of them can only work well under constrained conditions, such as all the faces are from a pre-defined closed set. In this paper, we propose a robust LSTM-Autoencoders (RLA) model to effectively restore partially occluded faces even in the wild. The RLA model consists of two LSTM components, which aims at occlusion-robust face encoding and recurrent occlusion removal respectively. The first one, named multi-scale spatial LSTM encoder, reads facial patches of various scales sequentially to output a latent representation, and occlusion-robustness is achieved owing to the fact that the influence of occlusion is only upon some of the patches. Receiving the representation learned by the encoder, the LSTM decoder with a dual channel architecture reconstructs the overall face and detects occlusion simultaneously, and by feat of LSTM, the decoder breaks down the task of face de-occlusion into restoring the occluded part step by step. Moreover, to minimize identify information loss and guarantee face recognition accuracy over recovered faces, we introduce an identity-preserving adversarial training scheme to further improve RLA. Extensive experiments on both synthetic and real datasets of faces with occlusion clearly demonstrate the effectiveness of our proposed RLA in removing different types of facial occlusion at various locations. The proposed method also provides significantly larger performance gain than other de-occlusion methods in promoting recognition performance over partially-occluded faces.

cs.CV↗

arXiv2016

Deep Edge Guided Recurrent Residual Learning for Image Super-Resolution

In this work, we consider the image super-resolution (SR) problem. The main challenge of image SR is to recover high-frequency details of a low-resolution (LR) image that are important for human perception. To address this essentially ill-posed problem, we introduce a Deep Edge Guided REcurrent rEsidual~(DEGREE) network to progressively recover the high-frequency details. Different from most of existing methods that aim at predicting high-resolution (HR) images directly, DEGREE investigates an alternative route to recover the difference between a pair of LR and HR images by recurrent residual learning. DEGREE further augments the SR process with edge-preserving capability, namely the LR image and its edge map can jointly infer the sharp edge details of the HR image during the recurrent recovery process. To speed up its training convergence rate, by-pass connections across multiple layers of DEGREE are constructed. In addition, we offer an understanding on DEGREE from the view-point of sub-band frequency decomposition on image signal and experimentally demonstrate how DEGREE can recover different frequency bands separately. Extensive experiments on three benchmark datasets clearly demonstrate the superiority of DEGREE over well-established baselines and DEGREE also provides new state-of-the-arts on these datasets.

cs.CV↗

arXiv2015

Deep Semantic Ranking Based Hashing for Multi-Label Image Retrieval

With the rapid growth of web images, hashing has received increasing interests in large scale image retrieval. Research efforts have been devoted to learning compact binary codes that preserve semantic similarity based on labels. However, most of these hashing methods are designed to handle simple binary similarity. The complex multilevel semantic structure of images associated with multiple labels have not yet been well explored. Here we propose a deep semantic ranking based method for learning hash functions that preserve multilevel semantic similarity between multi-label images. In our approach, deep convolutional neural network is incorporated into hash functions to jointly learn feature representations and mappings from them to hash codes, which avoids the limitation of semantic representation power of hand-crafted features. Meanwhile, a ranking list that encodes the multilevel similarity information is employed to guide the learning of such deep hash functions. An effective scheme based on surrogate loss is used to solve the intractable optimization problem of nonsmooth and multivariate ranking measures involved in the learning procedure. Experimental results show the superiority of our proposed approach over several state-of-the-art hashing methods in term of ranking evaluation metrics when tested on multi-label image datasets.

cs.CV↗

arXiv2015

Activity recognition for a smartphone and web based travel survey

In transport modeling and prediction, trip purposes play an important role since mobility choices (e.g. modes, routes, departure times) are made in order to carry out specific activities. Activity based models, which have been gaining popularity in recent years, are built from a large number of observed trips and their purposes. However, data acquired through traditional interview-based travel surveys lack the accuracy and quantity required by such models. Smartphones and interactive web interfaces have emerged as an attractive alternative to conventional travel surveys. A smartphone-based travel survey, Future Mobility Survey (FMS), was developed and field-tested in Singapore and collected travel data from more than 1000 participants for multiple days. To provide a more intelligent interface, inferring the activities of a user at a certain location is a crucial challenge. This paper presents a learning model that infers the most likely activity associated to a certain visited place. The data collected in FMS contain errors or noise due to various reasons, so a robust approach via ensemble learning is used to improve generalization performance. Our model takes advantage of cross-user historical data as well as user-specific information, including socio-demographics. Our empirical results using FMS data demonstrate that the proposed method contributes significantly to our travel survey application.

cs.CY↗

arXiv2012

A GPS Pseudorange Based Cooperative Vehicular Distance Measurement Technique

Accurate vehicular localization is important for various cooperative vehicle safety (CVS) applications such as collision avoidance, turning assistant, etc. In this paper, we propose a cooperative vehicular distance measurement technique based on the sharing of GPS pseudorange measurements and a weighted least squares method. The classic double difference pseudorange solution, which was originally designed for high-end survey level GPS systems, is adapted to low-end navigation level GPS receivers for its wide availability in ground vehicles. The Carrier to Noise Ratio (CNR) of raw pseudorange measurements are taken into account for noise mitigation. We present a Dedicated Short Range Communications (DSRC) based mechanism to implement the exchange of pseudorange information among neighboring vehicles. As demonstrated in field tests, our proposed technique increases the accuracy of the distance measurement significantly compared with the distance obtained from the GPS fixes.

cs.AI↗

arXiv2009

Network Coding for Multi-Resolution Multicast

Multi-resolution codes enable multicast at different rates to different receivers, a setup that is often desirable for graphics or video streaming. We propose a simple, distributed, two-stage message passing algorithm to generate network codes for single-source multicast of multi-resolution codes. The goal of this "pushback algorithm" is to maximize the total rate achieved by all receivers, while guaranteeing decodability of the base layer at each receiver. By conducting pushback and code generation stages, this algorithm takes advantage of inter-layer as well as intra-layer coding. Numerical simulations show that in terms of total rate achieved, the pushback algorithm outperforms routing and intra-layer coding schemes, even with codeword sizes as small as 10 bits. In addition, the performance gap widens as the number of receivers and the number of nodes in the network increases. We also observe that naiive inter-layer coding schemes may perform worse than intra-layer schemes under certain network conditions.

cs.NI↗

arXiv2009

On Counteracting Byzantine Attacks in Network Coded Peer-to-Peer Networks

Random linear network coding can be used in peer-to-peer networks to increase the efficiency of content distribution and distributed storage. However, these systems are particularly susceptible to Byzantine attacks. We quantify the impact of Byzantine attacks on the coded system by evaluating the probability that a receiver node fails to correctly recover a file. We show that even for a small probability of attack, the system fails with overwhelming probability. We then propose a novel signature scheme that allows packet-level Byzantine detection. This scheme allows one-hop containment of the contamination, and saves bandwidth by allowing nodes to detect and drop the contaminated packets. We compare the net cost of our signature scheme with various other Byzantine schemes, and show that when the probability of Byzantine attacks is high, our scheme is the most bandwidth efficient.

cs.NI↗

arXiv2006

Minimum-Cost Multicast over Coded Packet Networks

We consider the problem of establishing minimum-cost multicast connections over coded packet networks, i.e. packet networks where the contents of outgoing packets are arbitrary, causal functions of the contents of received packets. We consider both wireline and wireless packet networks as well as both static multicast (where membership of the multicast group remains constant for the duration of the connection) and dynamic multicast (where membership of the multicast group changes in time, with nodes joining and leaving the group). For static multicast, we reduce the problem to a polynomial-time solvable optimization problem, and we present decentralized algorithms for solving it. These algorithms, when coupled with existing decentralized schemes for constructing network codes, yield a fully decentralized approach for achieving minimum-cost multicast. By contrast, establishing minimum-cost static multicast connections over routed packet networks is a very difficult problem even using centralized computation, except in the special cases of unicast and broadcast connections. For dynamic multicast, we reduce the problem to a dynamic programming problem and apply the theory of dynamic programming to suggest how it may be solved.

cs.IT↗