arXiv subjects

Zhihui Wang

Publications and source records attributed to Zhihui Wang.

At least 73 records · Page 4Linked to original sources

arXiv2021

A New Dataset, Poisson GAN and AquaNet for Underwater Object Grabbing

To boost the object grabbing capability of underwater robots for open-sea farming, we propose a new dataset (UDD) consisting of three categories (seacucumber, seaurchin, and scallop) with 2,227 images. To the best of our knowledge, it is the first 4K HD dataset collected in a real open-sea farm. We also propose a novel Poisson-blending Generative Adversarial Network (Poisson GAN) and an efficient object detection network (AquaNet) to address two common issues within related datasets: the class-imbalance problem and the problem of mass small object, respectively. Specifically, Poisson GAN combines Poisson blending into its generator and employs a new loss called Dual Restriction loss (DR loss), which supervises both implicit space features and image-level features during training to generate more realistic images. By utilizing Poisson GAN, objects of minority class like seacucumber or scallop could be added into an image naturally and annotated automatically, which could increase the loss of minority classes during training detectors to eliminate the class-imbalance problem; AquaNet is a high-efficiency detector to address the problem of detecting mass small objects from cloudy underwater pictures. Within it, we design two efficient components: a depth-wise-convolution-based Multi-scale Contextual Features Fusion (MFF) block and a Multi-scale Blursampling (MBP) module to reduce the parameters of the network to 1.3 million. Both two components could provide multi-scale features of small objects under a short backbone configuration without any loss of accuracy. In addition, we construct a large-scale augmented dataset (AUDD) and a pre-training dataset via Poisson GAN from UDD. Extensive experiments show the effectiveness of the proposed Poisson GAN, AquaNet, UDD, AUDD, and pre-training dataset.

cs.CV↗

arXiv2021

2nd Place Solution for Waymo Open Dataset Challenge -- Real-time 2D Object Detection

In an autonomous driving system, it is essential to recognize vehicles, pedestrians and cyclists from images. Besides the high accuracy of the prediction, the requirement of real-time running brings new challenges for convolutional network models. In this report, we introduce a real-time method to detect the 2D objects from images. We aggregate several popular one-stage object detectors and train the models of variety input strategies independently, to yield better performance for accurate multi-scale detection of each category, especially for small objects. For model acceleration, we leverage TensorRT to optimize the inference time of our detection pipeline. As shown in the leaderboard, our proposed detection framework ranks the 2nd place with 75.00% L1 mAP and 69.72% L2 mAP in the real-time 2D detection track of the Waymo Open Dataset Challenges, while our framework achieves the latency of 45.8ms/frame on an Nvidia Tesla V100 GPU.

cs.CV↗

arXiv2021

A Dataset And Benchmark Of Underwater Object Detection For Robot Picking

Underwater object detection for robot picking has attracted a lot of interest. However, it is still an unsolved problem due to several challenges. We take steps towards making it more realistic by addressing the following challenges. Firstly, the currently available datasets basically lack the test set annotations, causing researchers must compare their method with other SOTAs on a self-divided test set (from the training set). Training other methods lead to an increase in workload and different researchers divide different datasets, resulting there is no unified benchmark to compare the performance of different algorithms. Secondly, these datasets also have other shortcomings, e.g., too many similar images or incomplete labels. Towards these challenges we introduce a dataset, Detecting Underwater Objects (DUO), and a corresponding benchmark, based on the collection and re-annotation of all relevant datasets. DUO contains a collection of diverse underwater images with more rational annotations. The corresponding benchmark provides indicators of both efficiency and accuracy of SOTAs (under the MMDtection framework) for academic research and industrial applications, where JETSON AGX XAVIER is used to assess detector speed to simulate the robot-embedded environment.

cs.CV↗

arXiv2021

Practical Verification of Quantum Properties in Quantum Approximate Optimization Runs

In order to assess whether quantum resources can provide an advantage over classical computation, it is necessary to characterize and benchmark the non-classical properties of quantum algorithms in a practical manner. In this paper, we show that using measurements in no more than 3 out of the possible $3^N$ bases, one can not only reconstruct the single-qubit reduced density matrices and measure the ability to create coherent superpositions, but also possibly verify entanglement across all $N$ qubits participating in the algorithm. We introduce a family of generalized Bell-type observables for which we establish an upper bound to the expectation values in fully separable states by proving a generalization of the Cauchy-Schwarz inequality, which may serve of independent interest. We demonstrate that a subset of such observables can serve as entanglement witnesses for QAOA-MaxCut states, and further argue that they are especially well tailored for this purpose by defining and computing an entanglement potency metric on witnesses. A subset of these observables also certify, in a weaker sense, the entanglement in GHZ states, which share the $\mathbb{Z}_2$ symmetry of QAOA-MaxCut. The construction of such witnesses follows directly from the cost Hamiltonian to be optimized, and not through the standard technique of using the projector of the state being certified. It may thus provide insights to construct similar witnesses for other variational algorithms prevalent in the NISQ era. We demonstrate our ideas with proof-of-concept experiments on the Rigetti Aspen-9 chip for ansatze containing up to 24 qubits.

quant-ph↗

arXiv2021

Accurate Monocular Object Detection via Color-Embedded 3D Reconstruction for Autonomous Driving

In this paper, we propose a monocular 3D object detection framework in the domain of autonomous driving. Unlike previous image-based methods which focus on RGB feature extracted from 2D images, our method solves this problem in the reconstructed 3D space in order to exploit 3D contexts explicitly. To this end, we first leverage a stand-alone module to transform the input data from 2D image plane to 3D point clouds space for a better input representation, then we perform the 3D detection using PointNet backbone net to obtain objects 3D locations, dimensions and orientations. To enhance the discriminative capability of point clouds, we propose a multi-modal feature fusion module to embed the complementary RGB cue into the generated point clouds representation. We argue that it is more effective to infer the 3D bounding boxes from the generated 3D scene space (i.e., X,Y, Z space) compared to the image plane (i.e., R,G,B image plane). Evaluation on the challenging KITTI dataset shows that our approach boosts the performance of state-of-the-art monocular approach by a large margin.

cs.CV↗

arXiv2021

Learning Scene Structure Guidance via Cross-Task Knowledge Transfer for Single Depth Super-Resolution

Existing color-guided depth super-resolution (DSR) approaches require paired RGB-D data as training samples where the RGB image is used as structural guidance to recover the degraded depth map due to their geometrical similarity. However, the paired data may be limited or expensive to be collected in actual testing environment. Therefore, we explore for the first time to learn the cross-modality knowledge at training stage, where both RGB and depth modalities are available, but test on the target dataset, where only single depth modality exists. Our key idea is to distill the knowledge of scene structural guidance from RGB modality to the single DSR task without changing its network architecture. Specifically, we construct an auxiliary depth estimation (DE) task that takes an RGB image as input to estimate a depth map, and train both DSR task and DE task collaboratively to boost the performance of DSR. Upon this, a cross-task interaction module is proposed to realize bilateral cross task knowledge transfer. First, we design a cross-task distillation scheme that encourages DSR and DE networks to learn from each other in a teacher-student role-exchanging fashion. Then, we advance a structure prediction (SP) task that provides extra structure regularization to help both DSR and DE networks learn more informative structure representations for depth recovery. Extensive experiments demonstrate that our scheme achieves superior performance in comparison with other DSR methods.

cs.CV↗

arXiv2021

Quality-Aware Network for Human Parsing

How to estimate the quality of the network output is an important issue, and currently there is no effective solution in the field of human parsing. In order to solve this problem, this work proposes a statistical method based on the output probability map to calculate the pixel quality information, which is called pixel score. In addition, the Quality-Aware Module (QAM) is proposed to fuse the different quality information, the purpose of which is to estimate the quality of human parsing results. We combine QAM with a concise and effective network design to propose Quality-Aware Network (QANet) for human parsing. Benefiting from the superiority of QAM and QANet, we achieve the best performance on three multiple and one single human parsing benchmarks, including CIHP, MHP-v2, Pascal-Person-Part and LIP. Without increasing the training and inference time, QAM improves the AP$^\text{r}$ criterion by more than 10 points in the multiple human parsing task. QAM can be extended to other tasks with good quality estimation, e.g. instance segmentation. Specifically, QAM improves Mask R-CNN by ~1% mAP on COCO and LVISv1.0 datasets. Based on the proposed QAM and QANet, our overall system wins 1st place in CVPR2019 COCO DensePose Challenge, and 1st place in Track 1 & 2 of CVPR2020 LIP Challenge. Code and models are available at https://github.com/soeaver/QANet.

cs.CV↗

arXiv2021

Character randomized benchmarking for non-multiplicity-free groups with applications to subspace, leakage, and matchgate randomized benchmarking

Randomized benchmarking (RB) is a powerful method for determining the error rate of experimental quantum gates. Traditional RB, however, is restricted to gatesets, such as the Clifford group, that form a unitary 2-design. The recently introduced character RB can benchmark more general gates using techniques from representation theory; up to now, however, this method has only been applied to "multiplicity-free" groups, a mathematical restriction on these groups. In this paper, we extend the original character RB derivation to explicitly treat non-multiplicity-free groups, and derive several applications. First, we derive a rigorous version of the recently introduced subspace RB, which seeks to characterize a set of one- and two-qubit gates that are symmetric under SWAP. Second, we develop a new leakage RB protocol that applies to more general groups of gates. Finally, we derive a scalable RB protocol for the matchgate group, a group that like the Clifford group is non-universal but becomes universal with the addition of one additional gate. This example provides one of the few examples of a scalable non-Clifford RB protocol. In all three cases, compared to existing theories, our method requires similar resources, but either provides a more accurate estimate of gate fidelity, or applies to a more general group of gates. In conclusion, we discuss the potential, and challenges, of using non-multiplicity-free character RB to develop new classes of scalable RB protocols and methods of characterizing specific gates.

quant-ph↗

arXiv2021

A Unified Joint Maximum Mean Discrepancy for Domain Adaptation

Domain adaptation has received a lot of attention in recent years, and many algorithms have been proposed with impressive progress. However, it is still not fully explored concerning the joint probability distribution (P(X, Y)) distance for this problem, since its empirical estimation derived from the maximum mean discrepancy (joint maximum mean discrepancy, JMMD) will involve complex tensor-product operator that is hard to manipulate. To solve this issue, this paper theoretically derives a unified form of JMMD that is easy to optimize, and proves that the marginal, class conditional and weighted class conditional probability distribution distances are our special cases with different label kernels, among which the weighted class conditional one not only can realize feature alignment across domains in the category level, but also deal with imbalance dataset using the class prior probabilities. From the revealed unified JMMD, we illustrate that JMMD degrades the feature-label dependence (discriminability) that benefits to classification, and it is sensitive to the label distribution shift when the label kernel is the weighted class conditional one. Therefore, we leverage Hilbert Schmidt independence criterion and propose a novel MMD matrix to promote the dependence, and devise a novel label kernel that is robust to label distribution shift. Finally, we conduct extensive experiments on several cross-domain datasets to demonstrate the validity and effectiveness of the revealed theoretical results.

cs.LG↗

arXiv2020

Rapid production of defect-free beryllium ion Coulomb crystals

Trapped atomic ions find wide applications ranging from precision measurement to quantum information science and quantum computing. Among the different atomic species employed, beryllium ions are widely used due to its light mass and convenient atomic structure. However, the ion loading process requires a high temperature for sufficient vapor pressure, generating undesirable gas load for the background vacuum and limiting the lifetime of a long ion chain. Here, we demonstrate a simple method to rapidly produce pure linear chains of beryllium ions with pulsed laser ablation, serving as a starting point for large-scale quantum information processing. Our method is fast compared to thermal ovens; reduces the vacuum load to only 10^(-12) Torr level; yields a short recovery time of a few seconds; and also eliminates the need for a deep ultraviolet laser for photo-ionization. In addition, we apply feedback control to obtain defect-free ion arrays with desirable lengths.

physics.atom-ph↗

arXiv2020

Quantum Approximate Optimization Algorithm for MaxCut: A Fermionic View

Farhi et al. recently proposed a class of quantum algorithms, the Quantum Approximate Optimization Algorithm (QAOA), for approximately solving combinatorial optimization problems. A level-p QAOA circuit consists of p steps; in each step a classical Hamiltonian, derived from the cost function, is applied followed by a mixing Hamiltonian. The 2p times for which these two Hamiltonians are applied are the parameters of the algorithm, which are to be optimized classically for the best performance. As p increases, parameter optimization becomes inefficient due to the curse of dimensionality. The success of the QAOA approach will depend, in part, on finding effective parameter-setting strategies. Here, we analytically and numerically study parameter setting for QAOA applied to MaxCut. For level-1 QAOA, we derive an analytical expression for a general graph. In principle, expressions for higher p could be derived, but the number of terms quickly becomes prohibitive. For a special case of MaxCut, the ring of disagrees, or the 1D antiferromagnetic ring, we provide an analysis for arbitrarily high level. Using a fermionic representation, the evolution of the system under QAOA translates into quantum control of an ensemble of independent spins. This treatment enables us to obtain analytical expressions for the performance of QAOA for any p. It also greatly simplifies numerical search for the optimal values of the parameters. By exploring symmetries, we identify a lower-dimensional sub-manifold of interest; the search effort can be accordingly reduced. This analysis also explains an observed symmetry in the optimal parameter values. Further, we numerically investigate the parameter landscape and show that it is a simple one in the sense of having no local optima.

quant-ph↗

arXiv2020

Direct Depth Learning Network for Stereo Matching

Being a crucial task of autonomous driving, Stereo matching has made great progress in recent years. Existing stereo matching methods estimate disparity instead of depth. They treat the disparity errors as the evaluation metric of the depth estimation errors, since the depth can be calculated from the disparity according to the triangulation principle. However, we find that the error of the depth depends not only on the error of the disparity but also on the depth range of the points. Therefore, even if the disparity error is low, the depth error is still large, especially for the distant points. In this paper, a novel Direct Depth Learning Network (DDL-Net) is designed for stereo matching. DDL-Net consists of two stages: the Coarse Depth Estimation stage and the Adaptive-Grained Depth Refinement stage, which are all supervised by depth instead of disparity. Specifically, Coarse Depth Estimation stage uniformly samples the matching candidates according to depth range to construct cost volume and output coarse depth. Adaptive-Grained Depth Refinement stage performs further matching near the coarse depth to correct the imprecise matching and wrong matching. To make the Adaptive-Grained Depth Refinement stage robust to the coarse depth and adaptive to the depth range of the points, the Granularity Uncertainty is introduced to Adaptive-Grained Depth Refinement stage. Granularity Uncertainty adjusts the matching range and selects the candidates' features according to coarse prediction confidence and depth range. We verify the performance of DDL-Net on SceneFlow dataset and DrivingStereo dataset by different depth metrics. Results show that DDL-Net achieves an average improvement of 25% on the SceneFlow dataset and $12\%$ on the DrivingStereo dataset comparing the classical methods. More importantly, we achieve state-of-the-art accuracy at a large distance.

cs.CV↗

arXiv2020

Full Matching on Low Resolution for Disparity Estimation

A Multistage Full Matching disparity estimation scheme (MFM) is proposed in this work. We demonstrate that decouple all similarity scores directly from the low-resolution 4D volume step by step instead of estimating low-resolution 3D cost volume through focusing on optimizing the low-resolution 4D volume iteratively leads to more accurate disparity. To this end, we first propose to decompose the full matching task into multiple stages of the cost aggregation module. Specifically, we decompose the high-resolution predicted results into multiple groups, and every stage of the newly designed cost aggregation module learns only to estimate the results for a group of points. This alleviates the problem of feature internal competitive when learning similarity scores of all candidates from one low-resolution 4D volume output from one stage. Then, we propose the strategy of \emph{Stages Mutual Aid}, which takes advantage of the relationship of multiple stages to boost similarity scores estimation of each stage, to solve the unbalanced prediction of multiple stages caused by serial multistage framework. Experiment results demonstrate that the proposed method achieves more accurate disparity estimation results and outperforms state-of-the-art methods on Scene Flow, KITTI 2012 and KITTI 2015 datasets.

cs.CV↗

arXiv2020

Quantum algorithms with local particle number conservation: noise effects and error correction

Quantum circuits with local particle number conservation (LPNC) restrict the quantum computation to a subspace of the Hilbert space of the qubit register. In a noiseless or fault-tolerant quantum computation, such quantities are preserved. In the presence of noise, however, the evolution's symmetry could be broken and non-valid states could be sampled at the end of the computation. On the other hand, the restriction to a subspace in the ideal case suggest the possibility of more resource efficient error mitigation techniques for circuits preserving symmetries that are not possible for general circuits. Here, we analyze the probability of staying in such symmetry-preserved subspaces under noise, providing an exact formula for local depolarizing noise. We apply our findings to benchmark, under depolarizing noise, the symmetry robustness of XY-QAOA, which has local particle number conserving symmetries, and is a special case of the Quantum Alternating Operator Ansatz. We also analyze the influence of the choice of encoding the problem on the symmetry robustness of the algorithm and discuss a simple adaption of the bit flip code to correct for symmetry-breaking errors with reduced resources.

quant-ph↗

arXiv2020

CPM R-CNN: Calibrating Point-guided Misalignment in Object Detection

In object detection, offset-guided and point-guided regression dominate anchor-based and anchor-free method separately. Recently, point-guided approach is introduced to anchor-based method. However, we observe points predicted by this way are misaligned with matched region of proposals and score of localization, causing a notable gap in performance. In this paper, we propose CPM R-CNN which contains three efficient modules to optimize anchor-based point-guided method. According to sufficient evaluations on the COCO dataset, CPM R-CNN is demonstrated efficient to improve the localization accuracy by calibrating mentioned misalignment. Compared with Faster R-CNN and Grid R-CNN based on ResNet-101 with FPN, our approach can substantially improve detection mAP by 3.3% and 1.5% respectively without whistles and bells. Moreover, our best model achieves improvement by a large margin to 49.9% on COCO test-dev. Code and models will be publicly available.

cs.CV↗

arXiv2020

Renovating Parsing R-CNN for Accurate Multiple Human Parsing

Multiple human parsing aims to segment various human parts and associate each part with the corresponding instance simultaneously. This is a very challenging task due to the diverse human appearance, semantic ambiguity of different body parts, and complex background. Through analysis of multiple human parsing task, we observe that human-centric global perception and accurate instance-level parsing scoring are crucial for obtaining high-quality results. But the most state-of-the-art methods have not paid enough attention to these issues. To reverse this phenomenon, we present Renovating Parsing R-CNN (RP R-CNN), which introduces a global semantic enhanced feature pyramid network and a parsing re-scoring network into the existing high-performance pipeline. The proposed RP R-CNN adopts global semantic representation to enhance multi-scale features for generating human parsing maps, and regresses a confidence score to represent its quality. Extensive experiments show that RP R-CNN performs favorably against state-of-the-art methods on CIHP and MHP-v2 datasets. Code and models are available at https://github.com/soeaver/RP-R-CNN.

cs.CV↗

arXiv2020

Rethink Maximum Mean Discrepancy for Domain Adaptation

Existing domain adaptation methods aim to reduce the distributional difference between the source and target domains and respect their specific discriminative information, by establishing the Maximum Mean Discrepancy (MMD) and the discriminative distances. However, they usually accumulate to consider those statistics and deal with their relationships by estimating parameters blindly. This paper theoretically proves two essential facts: 1) minimizing the MMD equals to maximize the source and target intra-class distances respectively but jointly minimize their variance with some implicit weights, so that the feature discriminability degrades; 2) the relationship between the intra-class and inter-class distances is as one falls, another rises. Based on this, we propose a novel discriminative MMD. On one hand, we consider the intra-class and inter-class distances alone to remove a redundant parameter, and the revealed weights provide their approximate optimal ranges. On the other hand, we design two different strategies to boost the feature discriminability: 1) we directly impose a trade-off parameter on the implicit intra-class distance in MMD to regulate its change; 2) we impose the similar weights revealed in MMD on inter-class distance and maximize it, then a balanced factor could be introduced to quantitatively leverage the relative importance between the feature transferability and its discriminability. The experiments on several benchmark datasets not only prove the validity of theoretical results but also demonstrate that our approach could perform better than the comparative state-of-art methods substantially.

cs.LG↗

arXiv2020

Ferromagnetically shifting the power of pausing

We study the interplay between quantum annealing parameters in embedded problems, providing both deeper insights into the physics of these devices and pragmatic recommendations to improve performance on optimization problems. We choose as our test case the class of degree-bounded minimum spanning tree problems. Through runs on a D-Wave quantum annealer, we demonstrate that pausing in a specific time window in the anneal provides improvement in the probability of success and in the time-to-solution for these problems. The time window is consistent across problem instances, and its location is within the region suggested by prior theory and seen in previous results on native problems. An approach to enable gauge transformations for problems with the qubit coupling strength $J$ in an asymmetric range is presented and shown to significantly improve performance. We also confirm that the optimal pause location exhibits a shift with the magnitude of the ferromagnetic coupling, $|J_F|$, between physical qubits representing the same logical one. We extend the theoretical picture for pausing and thermalization in quantum annealing to the embedded case. This picture, along with perturbation theory analysis, and exact numerical results on small problems, confirms that the effective pause region moves earlier in the anneal as $|J_F|$ increases. It also suggests why pausing, while still providing significant benefit, has a less pronounced effect on embedded problems.

quant-ph↗