arXiv subjects

Jiahui Huang

Publications and source records attributed to Jiahui Huang.

79 records · Page 5Linked to original sources

arXiv2020

ClusterVO: Clustering Moving Instances and Estimating Visual Odometry for Self and Surroundings

We present ClusterVO, a stereo Visual Odometry which simultaneously clusters and estimates the motion of both ego and surrounding rigid clusters/objects. Unlike previous solutions relying on batch input or imposing priors on scene structure or dynamic object models, ClusterVO is online, general and thus can be used in various scenarios including indoor scene understanding and autonomous driving. At the core of our system lies a multi-level probabilistic association mechanism and a heterogeneous Conditional Random Field (CRF) clustering approach combining semantic, spatial and motion information to jointly infer cluster segmentations online for every frame. The poses of camera and dynamic objects are instantly solved through a sliding-window optimization. Our system is evaluated on Oxford Multimotion and KITTI dataset both quantitatively and qualitatively, reaching comparable results to state-of-the-art solutions on both odometry and dynamic trajectory recovery.

cs.CV↗

arXiv2020

Shallow2Deep: Indoor Scene Modeling by Single Image Understanding

Dense indoor scene modeling from 2D images has been bottlenecked due to the absence of depth information and cluttered occlusions. We present an automatic indoor scene modeling approach using deep features from neural networks. Given a single RGB image, our method simultaneously recovers semantic contents, 3D geometry and object relationship by reasoning indoor environment context. Particularly, we design a shallow-to-deep architecture on the basis of convolutional networks for semantic scene understanding and modeling. It involves multi-level convolutional networks to parse indoor semantics/geometry into non-relational and relational knowledge. Non-relational knowledge extracted from shallow-end networks (e.g. room layout, object geometry) is fed forward into deeper levels to parse relational semantics (e.g. support relationship). A Relation Network is proposed to infer the support relationship between objects. All the structured semantics and geometry above are assembled to guide a global optimization for 3D scene modeling. Qualitative and quantitative analysis demonstrates the feasibility of our method in understanding and modeling semantics-enriched indoor scenes by evaluating the performance of reconstruction accuracy, computation performance and scene complexity.

cs.CV↗

arXiv2020

What fraction of an $S_n$-orbit can lie on a hyperplane?

Consider the $S_n$-action on $\mathbb{R}^n$ given by permuting coordinates. This paper addresses the following problem: compute $\max_{v,H} |H\cap S_nv|$ as $H\subset\mathbb{R}^n$ ranges over all hyperplanes through the origin and $v\in\mathbb{R}^n$ ranges over all vectors with distinct coordinates that are not contained in the hyperplane $\sum x_i=0$. We conjecture that for $n\geq3$, the answer is $(n-1)!$ for odd $n$, and $n(n-2)!$ for even $n$. We prove that if $p$ is the largest prime with $p\leq n$, then $\max_{v,H} |H\cap S_nv|\leq \frac{n!}{p}$. In particular, this proves the conjecture when $n$ or $n-1$ is prime.

math.CO↗

arXiv2019

Deep Anchored Convolutional Neural Networks

Convolutional Neural Networks (CNNs) have been proven to be extremely successful at solving computer vision tasks. State-of-the-art methods favor such deep network architectures for its accuracy performance, with the cost of having massive number of parameters and high weights redundancy. Previous works have studied how to prune such CNNs weights. In this paper, we go to another extreme and analyze the performance of a network stacked with a single convolution kernel across layers, as well as other weights sharing techniques. We name it Deep Anchored Convolutional Neural Network (DACNN). Sharing the same kernel weights across layers allows to reduce the model size tremendously, more precisely, the network is compressed in memory by a factor of L, where L is the desired depth of the network, disregarding the fully connected layer for prediction. The number of parameters in DACNN barely increases as the network grows deeper, which allows us to build deep DACNNs without any concern about memory costs. We also introduce a partial shared weights network (DACNN-mix) as well as an easy-plug-in module, coined regulators, to boost the performance of our architecture. We validated our idea on 3 datasets: CIFAR-10, CIFAR-100 and SVHN. Our results show that we can save massive amounts of memory with our model, while maintaining a high accuracy performance.

cs.CV↗

arXiv2019

PolarLight: a CubeSat X-ray Polarimeter based on the Gas Pixel Detector

The gas pixel detector (GPD) is designed and developed for high-sensitivity astronomical X-ray polarimetry, which is a new window about to open in a few years. Due to the small mass, low power, and compact geometry of the GPD, we propose a CubeSat mission Polarimeter Light (PolarLight) to demonstrate and test the technology directly in space. There is no optics but a collimator to constrain the field of view to 2.3 degrees. Filled with pure dimethyl ether (DME) at 0.8 atm and sealed by a beryllium window of 100 micron thick, with a sensitive area of about 1.4 mm by 1.4 mm, PolarLight allows us to observe the brightest X-ray sources on the sky, with a count rate of, e.g., ~0.2 counts/s from the Crab nebula. The PolarLight is 1U in size and mounted in a 6U CubeSat, which was launched into a low Earth Sun-synchronous orbit on October 29, 2018, and is currently under test. More launches with improved designs are planned in 2019. These tests will help increase the technology readiness for future missions such as the enhanced X-ray Timing and Polarimetry (eXTP), better understand the orbital background, and may help constrain the physics with observations of the brightest objects.

astro-ph.IM↗

arXiv2019

DeepSpline: Data-Driven Reconstruction of Parametric Curves and Surfaces

Reconstruction of geometry based on different input modes, such as images or point clouds, has been instrumental in the development of computer aided design and computer graphics. Optimal implementations of these applications have traditionally involved the use of spline-based representations at their core. Most such methods attempt to solve optimization problems that minimize an output-target mismatch. However, these optimization techniques require an initialization that is close enough, as they are local methods by nature. We propose a deep learning architecture that adapts to perform spline fitting tasks accordingly, providing complementary results to the aforementioned traditional methods. We showcase the performance of our approach, by reconstructing spline curves and surfaces based on input images or point clouds.

cs.CV↗

arXiv2018

Enhanced interlayer neutral excitons and trions in trilayer van der Waals heterostructures

Vertically stacked van der Waals heterostructures constitute a promising platform for providing tailored band alignment with enhanced excitonic systems. Here we report observations of neutral and charged interlayer excitons in trilayer WSe2-MoSe2-WSe2 van der Waals heterostructures and their dynamics. The addition of a WSe2 layer in the trilayer leads to significantly higher photoluminescence quantum yields and tunable spectral resonance compared to its bilayer heterostructures at cryogenic temperatures. The observed enhancement in the photoluminescence quantum yield is due to significantly larger electron-hole overlap and higher light absorbance in the trilayer heterostructure, supported via first-principle pseudopotential calculations based on spin-polarized density functional theory. We further uncover the temperature- and power-dependence, as well as time-resolved photoluminescence of the trilayer heterostructure interlayer neutral excitons and trions. Our study elucidates the prospects of manipulating light emission from interlayer excitons and designing atomic heterostructures from first-principles for optoelectronics.

cond-mat.mes-hall↗