Searcharxiv⌕ Search

arXiv subjects

Qi Feng

Publications and source records attributed to Qi Feng.

At least 91 records · Page 5Linked to original sources
arXiv2021

The 5th AI City Challenge

The AI City Challenge was created with two goals in mind: (1) pushing the boundaries of research and development in intelligent video analysis for smarter cities use cases, and (2) assessing tasks where the level of performance is enough to cause real-world adoption. Transportation is a segment ripe for such adoption. The fifth AI City Challenge attracted 305 participating teams across 38 countries, who leveraged city-scale real traffic data and high-quality synthetic data to compete in five challenge tracks. Track 1 addressed video-based automatic vehicle counting, where the evaluation being conducted on both algorithmic effectiveness and computational efficiency. Track 2 addressed city-scale vehicle re-identification with augmented synthetic data to substantially increase the training set for the task. Track 3 addressed city-scale multi-target multi-camera vehicle tracking. Track 4 addressed traffic anomaly detection. Track 5 was a new track addressing vehicle retrieval using natural language descriptions. The evaluation system shows a general leader board of all submitted results, and a public leader board of results limited to the contest participation rules, where teams are not allowed to use external data in their work. The public leader board shows results more close to real-world situations where annotated data is limited. Results show the promise of AI in Smarter Transportation. State-of-the-art performance for some tasks shows that these technologies are ready for adoption in real-world systems.

cs.CV↗
arXiv2021

Siamese Natural Language Tracker: Tracking by Natural Language Descriptions with Siamese Trackers

We propose a novel Siamese Natural Language Tracker (SNLT), which brings the advancements in visual tracking to the tracking by natural language (NL) descriptions task. The proposed SNLT is applicable to a wide range of Siamese trackers, providing a new class of baselines for the tracking by NL task and promising future improvements from the advancements of Siamese trackers. The carefully designed architecture of the Siamese Natural Language Region Proposal Network (SNL-RPN), together with the Dynamic Aggregation of vision and language modalities, is introduced to perform the tracking by NL task. Empirical results over tracking benchmarks with NL annotations show that the proposed SNLT improves Siamese trackers by 3 to 7 percentage points with a slight tradeoff of speed. The proposed SNLT outperforms all NL trackers to-date and is competitive among state-of-the-art real-time trackers on LaSOT benchmarks while running at 50 frames per second on a single GPU.

cs.CV↗
arXiv2021

CityFlow-NL: Tracking and Retrieval of Vehicles at City Scale by Natural Language Descriptions

Natural Language (NL) descriptions can be one of the most convenient or the only way to interact with systems built to understand and detect city scale traffic patterns and vehicle-related events. In this paper, we extend the widely adopted CityFlow Benchmark with NL descriptions for vehicle targets and introduce the CityFlow-NL Benchmark. The CityFlow-NL contains more than 5,000 unique and precise NL descriptions of vehicle targets, making it the first multi-target multi-camera tracking with NL descriptions dataset to our knowledge. Moreover, the dataset facilitates research at the intersection of multi-object tracking, retrieval by NL descriptions, and temporal localization of events. In this paper, we focus on two foundational tasks: the Vehicle Retrieval by NL task and the Vehicle Tracking by NL task, which take advantage of the proposed CityFlow-NL benchmark and provide a strong basis for future research on the multi-target multi-camera tracking by NL description task.

cs.CV↗
arXiv2021

Non-convex Learning via Replica Exchange Stochastic Gradient MCMC

Replica exchange Monte Carlo (reMC), also known as parallel tempering, is an important technique for accelerating the convergence of the conventional Markov Chain Monte Carlo (MCMC) algorithms. However, such a method requires the evaluation of the energy function based on the full dataset and is not scalable to big data. The naïve implementation of reMC in mini-batch settings introduces large biases, which cannot be directly extended to the stochastic gradient MCMC (SGMCMC), the standard sampling method for simulating from deep neural networks (DNNs). In this paper, we propose an adaptive replica exchange SGMCMC (reSGMCMC) to automatically correct the bias and study the corresponding properties. The analysis implies an acceleration-accuracy trade-off in the numerical discretization of a Markov jump process in a stochastic environment. Empirically, we test the algorithm through extensive experiments on various setups and obtain the state-of-the-art results on CIFAR10, CIFAR100, and SVHN in both supervised learning and semi-supervised learning tasks.

stat.ML↗
arXiv2021

CE-FPN: Enhancing Channel Information for Object Detection

Feature pyramid network (FPN) has been an effective framework to extract multi-scale features in object detection. However, current FPN-based methods mostly suffer from the intrinsic flaw of channel reduction, which brings about the loss of semantical information. And the miscellaneous fused feature maps may cause serious aliasing effects. In this paper, we present a novel channel enhancement feature pyramid network (CE-FPN) with three simple yet effective modules to alleviate these problems. Specifically, inspired by sub-pixel convolution, we propose a sub-pixel skip fusion method to perform both channel enhancement and upsampling. Instead of the original 1x1 convolution and linear upsampling, it mitigates the information loss due to channel reduction. Then we propose a sub-pixel context enhancement module for extracting more feature representations, which is superior to other context methods due to the utilization of rich channel information by sub-pixel convolution. Furthermore, a channel attention guided module is introduced to optimize the final integrated features on each level, which alleviates the aliasing effect only with a few computational burdens. Our experiments show that CE-FPN achieves competitive performance compared to state-of-the-art FPN-based detectors on MS COCO benchmark.

cs.CV↗
arXiv2021

Accelerating Convergence of Replica Exchange Stochastic Gradient MCMC via Variance Reduction

Replica exchange stochastic gradient Langevin dynamics (reSGLD) has shown promise in accelerating the convergence in non-convex learning; however, an excessively large correction for avoiding biases from noisy energy estimators has limited the potential of the acceleration. To address this issue, we study the variance reduction for noisy energy estimators, which promotes much more effective swaps. Theoretically, we provide a non-asymptotic analysis on the exponential acceleration for the underlying continuous-time Markov jump process; moreover, we consider a generalized Girsanov theorem which includes the change of Poisson measure to overcome the crude discretization based on the Gröwall's inequality and yields a much tighter error in the 2-Wasserstein ($\mathcal{W}_2$) distance. Numerically, we conduct extensive experiments and obtain the state-of-the-art results in optimization and uncertainty estimates for synthetic experiments and image data.

stat.ML↗
arXiv2021

Dynamic Curves for Decentralized Autonomous Cryptocurrency Exchanges

One of the exciting recent developments in decentralized finance (DeFi) has been the development of decentralized cryptocurrency exchanges that can autonomously handle conversion between different cryptocurrencies. Decentralized exchange protocols such as Uniswap, Curve and other types of Automated Market Makers (AMMs) maintain a liquidity pool (LP) of two or more assets constrained to maintain at all times a mathematical relation to each other, defined by a given function or curve. Examples of such functions are the constant-sum and constant-product AMMs. Existing systems however suffer from several challenges. They require external arbitrageurs to restore the price of tokens in the pool to match the market price. Such activities can potentially drain resources from the liquidity pool. In particular, dramatic market price changes can result in low liquidity with respect to one or more of the assets and reduce the total value of the LP. We propose in this work a new approach to constructing the AMM by proposing the idea of dynamic curves. It utilizes input from a market price oracle to modify the mathematical relationship between the assets so that the pool price continuously and automatically adjusts to be identical to the market price. This approach eliminates arbitrage opportunities and, as we show through simulations, maintains liquidity in the LP for all assets and the total value of the LP over a wide range of market prices.

q-fin.TR↗
arXiv2020

Entropy dissipation via Information Gamma calculus: Non-reversible stochastic differential equations

We formulate explicit bounds to guarantee the exponential dissipation for some non-gradient stochastic differential equations towards their invariant distributions. Our method extends the connection between Gamma calculus and Hessian operators in $L^2$--Wasserstein space. In details, we apply Lyapunov methods in the space of probabilities, where the Lyapunov functional is chosen as the relative Fisher information. We derive the Fisher information induced Gamma calculus to handle non-gradient drift vector fields. We obtain the explicit dissipation bound in terms of $L_1$ distance and formulate the non-reversible Poincar{é} inequality. An analytical example is provided for a non-reversible Langevin dynamic.

math.PR↗
arXiv2020

Learning to Separate: Detecting Heavily-Occluded Objects in Urban Scenes

While visual object detection with deep learning has received much attention in the past decade, cases when heavy intra-class occlusions occur have not been studied thoroughly. In this work, we propose a Non-Maximum-Suppression (NMS) algorithm that dramatically improves the detection recall while maintaining high precision in scenes with heavy occlusions. Our NMS algorithm is derived from a novel embedding mechanism, in which the semantic and geometric features of the detected boxes are jointly exploited. The embedding makes it possible to determine whether two heavily-overlapping boxes belong to the same object in the physical world. Our approach is particularly useful for car detection and pedestrian detection in urban scenes where occlusions often happen. We show the effectiveness of our approach by creating a model called SG-Det (short for Semantics and Geometry Detection) and testing SG-Det on two widely-adopted datasets, KITTI and CityPersons for which it achieves state-of-the-art performance.

cs.CV↗
arXiv2020

Taylor expansions and Castell estimates for solutions of stochastic differential equations driven by rough paths

We study the Taylor expansion for the solutions of differential equations driven by $p$-rough paths with $p>2$. We prove a general theorem concerning the convergence of the Taylor expansion on a nonempty interval provided that the vector fields are analytic on a ball centered at the initial point. We also derive criteria that enable us to study the rate of convergence of the Taylor expansion. Finally and this is also the main and the most original part of this paper, we prove Castell expansions and tail estimates with exponential decays for the remainder terms of the solutions of the stochastic differential equations driven by continuous centered Gaussian process with finite $2D~ρ-$variation and fractional Brownian motion with Hurst parameter $H>1/4$.

math.PR↗
arXiv2020

Multi-wavelength observations of 2HWC J1928+177: dark accelerator or new TeV gamma-ray binary?

2HWC J1928+177 is a Galactic TeV gamma-ray source detected by the High Altitude Water Cherenkov (HAWC) Observatory up to ~ 56 TeV. The HAWC source, later confirmed by H.E.S.S., still remains unidentified as a dark accelerator since there is no apparent supernova remnant or pulsar wind nebula detected in the lower energy bands. The radio pulsar PSR J1928+1746, coinciding with the HAWC source position, has no X-ray counterpart. Our SED modeling shows that inverse Compton scattering in the putative pulsar wind nebula can account for the TeV emission only if the unseen nebula is extended beyond r ~ 4 [arcmin]. Alternatively, TeV gamma rays may be produced by hadronic interactions between relativistic protons from an undetected supernova remnant associated with the radio pulsar and a nearby molecular cloud G52.9+0.1. NuSTAR and Chandra observations detected a variable X-ray point source within the HAWC error circle, potentially associated with a bright IR source. The X-ray spectra can be fitted with an absorbed power-law model with $N_{\rm H} = (9\pm3)\times10^{22}$ cm$^{-2}$ and $Γ_X = 1.6\pm0.3$ and exhibit long-term X-ray flux variability over the last decade. If the X-ray source, possibly associated with the IR source (likely an O star), is the counterpart of the HAWC source, it may be a new TeV gamma-ray binary powered by collisions between the pulsar wind and stellar wind. Follow-up X-ray observations are warranted to search for diffuse X-ray emission and determine the nature of the HAWC source.

astro-ph.HE↗
arXiv2020

Investigating a Deep Learning Method to Analyze Images from Multiple Gamma-ray Telescopes

Imaging atmospheric Cherenkov telescope (IACT) arrays record images from air showers initiated by gamma rays entering the atmosphere, allowing astrophysical sources to be observed at very high energies. To maximize IACT sensitivity, gamma-ray showers must be efficiently distinguished from the dominant background of cosmic-ray showers using images from multiple telescopes. A combination of convolutional neural networks (CNNs) with a recurrent neural network (RNN) has been proposed to perform this task. Using CTLearn, an open source Python package using deep learning to analyze data from IACTs, with simulated data from the upcoming Cherenkov Telescope Array (CTA), we implement a CNN-RNN network and find no evidence that sorting telescope images by total amplitude improves background rejection performance.

astro-ph.IM↗
arXiv2019

Real-time Visual Object Tracking with Natural Language Description

In recent years, deep-learning-based visual object trackers have been studied thoroughly, but handling occlusions and/or rapid motion of the target remains challenging. In this work, we argue that conditioning on the natural language (NL) description of a target provides information for longer-term invariance, and thus helps cope with typical tracking challenges. However, deriving a formulation to combine the strengths of appearance-based tracking with the language modality is not straightforward. We propose a novel deep tracking-by-detection formulation that can take advantage of NL descriptions. Regions that are related to the given NL description are generated by a proposal network during the detection phase of the tracker. Our LSTM based tracker then predicts the update of the target from regions proposed by the NL based detection phase. In benchmarks, our method is competitive with state of the art trackers, while it outperforms all other trackers on targets with unambiguous and precise language annotations. It also beats the state-of-the-art NL tracker when initializing without a bounding box. Our method runs at over 30 fps on a single GPU.

cs.CV↗
arXiv2019

Density of the signature process of fBm

We study the density of the signature of fractional Brownian motions with parameter $H>1/4$. In particular, we prove existence, smoothness, global Gaussian upper bounds and Varadhan's type asymptotics for this density. A key result is that the estimates on the density we obtain are controlled by the Carnot-Carathéodory distance of the group.

math.PR↗
arXiv2018

Enhanced magnetic properties in ZnCoAlO caused by exchangecoupling to Co nanoparticles

Wereport the results of a sequence of magnetisation and magneto-optical studies on laser ablated thin films of ZnCoAlO and ZnCoO that contain a small amount of metallic cobalt. The results are compared to those expected when all the magnetization is due to isolated metallic clusters of cobalt and with an oxide sample that is almost free from metallic inclusions. Using a variety of direct magnetic measurements and also magnetic circular dichroism we find that there is ferromagnetism within both the oxide and the metallic inclusions, and furthermore that these magnetic components are exchange-coupled when aluminium is included. This enhances both the coercive field and the remanence. Hence the presence of a controlled quantity of metallic nanoparticles in ZnAlO can improve the magnetic response of the oxide, thus giving great advantages for applications in spintronics.

cond-mat.mtrl-sci↗
arXiv2018

Enhanced optical Kerr nonlinearity of graphene/Si hybrid waveguide

In this work, we experimentally study the optical kerr nonlinearities of graphene/Si hybrid waveguides with enhanced self-phase modulation. In the case of CMOS compatible materials for nonlinear optical signal processing, Si and silicon nitride waveguides have been extensively investigated over the past decade. However, Si waveguides exhibit strong two-photon absorption (TPA) at telecommunication wavelengths, which lead to a significant reduction of nonlinear figure of merit. In contrast, silicon nitride based material system usually suppress the TPA, but simultaneously leads to the reduction of the Kerr nonlinearity by two orders of magnitude. Here, we introduce a graphene/Si hybrid waveguide, which remain the optical properties and CMOS compatibility of Si waveguides, while enhance the Kerr nonlinearity by transferring patterned graphene over the top of the waveguides. The graphene/Si waveguides are measured with a nonlinear parameter of 510 W-1m-1. Enhanced nonlinear figure-of-merit (FOM) of 2.48 has been achieved, which is three times higher than that of the Si waveguide. This work reveals the potential application of graphene/Si hybrid photonic waveguides with high Kerr nonlinearity and FOM for nonlinear all-optical signal processing.

physics.app-ph↗
arXiv2018

Harnack inequalities on totally geodesic foliations with transverse Ricci flow

In the current paper,under the transverse Ricci flow on a totally geodesic Riemannian foliation, we prove two types of differential Harnack inequalities (Li-Yau gradient estimate) for the positive solutions of the heat equation associated with the time dependent horizontal Laplacian operators. We also get a time dependent version of the generalized curvature dimension inequality. As a consequence of aforementioned results, we also get parabolic Harnack inequalities and heat kernel upper bounds.

math.DG↗