arXiv subjects

Guang Chen

Publications and source records attributed to Guang Chen.

At least 163 records · Page 9Linked to original sources

arXiv2020

Efficient Pig Counting in Crowds with Keypoints Tracking and Spatial-aware Temporal Response Filtering

Pig counting is a crucial task for large-scale pig farming, which is usually completed by human visually. But this process is very time-consuming and error-prone. Few studies in literature developed automated pig counting method. Existing methods only focused on pig counting using single image, and its accuracy is challenged by several factors, including pig movements, occlusion and overlapping. Especially, the field of view of a single image is very limited, and could not meet the requirements of pig counting for large pig grouping houses. To that end, we presented a real-time automated pig counting system in crowds using only one monocular fisheye camera with an inspection robot. Our system showed that it produces accurate results surpassing human. Our pipeline began with a novel bottom-up pig detection algorithm to avoid false negatives due to overlapping, occlusion and deformation of pigs. A deep convolution neural network (CNN) is designed to detect keypoints of pig body part and associate the keypoints to identify individual pigs. After that, an efficient on-line tracking method is used to associate pigs across video frames. Finally, a novel spatial-aware temporal response filtering (STRF) method is proposed to predict the counts of pigs, which is effective to suppress false positives caused by pig or camera movements or tracking failures. The whole pipeline has been deployed in an edge computing device, and demonstrated the effectiveness.

cs.CV↗

arXiv2020

Globally optimal vertical direction estimation in Atlanta World

In man-made environments, such as indoor and urban scenes, most of the objects and structures are organized in the form of orthogonal and parallel planes. These planes can be approximated by the Atlanta world assumption, in which the normals of planes can be represented by the Atlanta frames. Atlanta world assumption, which can be considered as a generalized Manhattan world assumption, has one vertical frame and multiple horizontal frames. Conventionally, given a set of inputs such as surface normals, the Atlanta frame estimation problem can be solved in one-time by branch-and-bound (BnB). However, the runtime of the BnB algorithm will increase greatly when the dimensionality (i.e., the number of horizontal frames) increases. In this paper, we estimate only the vertical direction instead of all Atlanta frames at once. Accordingly, we propose a vertical direction estimation method by considering the relationship between the vertical frame and horizontal frames. Concretely, our approach employs a BnB algorithm to search the vertical direction guaranteeing global optimality without requiring prior knowledge of the number of Atlanta frames. Four novel bounds by mapping 3D-hemisphere to a 2D region are investigated to guarantee convergence. We verify the validity of the proposed method in various challenging synthetic and real-world data.

cs.CV↗

arXiv2020

Gen-LaneNet: A Generalized and Scalable Approach for 3D Lane Detection

We present a generalized and scalable method, called Gen-LaneNet, to detect 3D lanes from a single image. The method, inspired by the latest state-of-the-art 3D-LaneNet, is a unified framework solving image encoding, spatial transform of features and 3D lane prediction in a single network. However, we propose unique designs for Gen-LaneNet in two folds. First, we introduce a new geometry-guided lane anchor representation in a new coordinate frame and apply a specific geometric transformation to directly calculate real 3D lane points from the network output. We demonstrate that aligning the lane points with the underlying top-view features in the new coordinate frame is critical towards a generalized method in handling unfamiliar scenes. Second, we present a scalable two-stage framework that decouples the learning of image segmentation subnetwork and geometry encoding subnetwork. Compared to 3D-LaneNet, the proposed Gen-LaneNet drastically reduces the amount of 3D lane labels required to achieve a robust solution in real-world application. Moreover, we release a new synthetic dataset and its construction strategy to encourage the development and evaluation of 3D lane detection methods. In experiments, we conduct extensive ablation study to substantiate the proposed Gen-LaneNet significantly outperforms 3D-LaneNet in average precision(AP) and F-score.

cs.CV↗

arXiv2020

Indirect and Direct Training of Spiking Neural Networks for End-to-End Control of a Lane-Keeping Vehicle

Building spiking neural networks (SNNs) based on biological synaptic plasticities holds a promising potential for accomplishing fast and energy-efficient computing, which is beneficial to mobile robotic applications. However, the implementations of SNNs in robotic fields are limited due to the lack of practical training methods. In this paper, we therefore introduce both indirect and direct end-to-end training methods of SNNs for a lane-keeping vehicle. First, we adopt a policy learned using the \textcolor{black}{Deep Q-Learning} (DQN) algorithm and then subsequently transfer it to an SNN using supervised learning. Second, we adopt the reward-modulated spike-timing-dependent plasticity (R-STDP) for training SNNs directly, since it combines the advantages of both reinforcement learning and the well-known spike-timing-dependent plasticity (STDP). We examine the proposed approaches in three scenarios in which a robot is controlled to keep within lane markings by using an event-based neuromorphic vision sensor. We further demonstrate the advantages of the R-STDP approach in terms of the lateral localization accuracy and training time steps by comparing them with other three algorithms presented in this paper.

cs.NE↗

arXiv2019

Neuromorphic Visual Odometry System for Intelligent Vehicle Application with Bio-inspired Vision Sensor

The neuromorphic camera is a brand new vision sensor that has emerged in recent years. In contrast to the conventional frame-based camera, the neuromorphic camera only transmits local pixel-level changes at the time of its occurrence and provides an asynchronous event stream with low latency. It has the advantages of extremely low signal delay, low transmission bandwidth requirements, rich information of edges, high dynamic range etc., which make it a promising sensor in the application of in-vehicle visual odometry system. This paper proposes a neuromorphic in-vehicle visual odometry system using feature tracking algorithm. To the best of our knowledge, this is the first in-vehicle visual odometry system that only uses a neuromorphic camera, and its performance test is carried out on actual driving datasets. In addition, an in-depth analysis of the results of the experiment is provided. The work of this paper verifies the feasibility of in-vehicle visual odometry system using neuromorphic cameras.

cs.RO↗

arXiv2019

A Novel Method for the Absolute Pose Problem with Pairwise Constraints

Absolute pose estimation is a fundamental problem in computer vision, and it is a typical parameter estimation problem, meaning that efforts to solve it will always suffer from outlier-contaminated data. Conventionally, for a fixed dimensionality d and the number of measurements N, a robust estimation problem cannot be solved faster than O(N^d). Furthermore, it is almost impossible to remove d from the exponent of the runtime of a globally optimal algorithm. However, absolute pose estimation is a geometric parameter estimation problem, and thus has special constraints. In this paper, we consider pairwise constraints and propose a globally optimal algorithm for solving the absolute pose estimation problem. The proposed algorithm has a linear complexity in the number of correspondences at a given outlier ratio. Concretely, we first decouple the rotation and the translation subproblems by utilizing the pairwise constraints, and then we solve the rotation subproblem using the branch-and-bound algorithm. Lastly, we estimate the translation based on the known rotation by using another branch-and-bound algorithm. The advantages of our method are demonstrated via thorough testing on both synthetic and real-world data

cs.CV↗

arXiv2018

Self-Localization of Parking Robots Using Square-Like Landmarks

In this paper, we present a framework for self-localization of parking robots in a parking lot innovatively using square-like landmarks, aiming to provide a positioning solution with low cost but high accuracy. It utilizes square structures common in parking lots such as pillars, corners or charging piles as robust landmarks and deduces the global pose of the robot in conjunction with an off-line map. The localization is performed in real-time via Particle Filter using a single line scanning LiDAR as main sensor, an odometry as secondary information sources. The system has been tested in a simulation environment built in V-REP, the result of which demonstrates its positioning accuracy below 0.20 m and a corresponding heading error below 1°.

cs.RO↗

arXiv2018

An Efficient L-Shape Fitting Method for Vehicle Pose Detection with 2D LiDAR

Detecting vehicles with strong robustness and high efficiency has become one of the key capabilities of fully autonomous driving cars. This topic has already been widely studied by GPU-accelerated deep learning approaches using image sensors and 3D LiDAR, however, few studies seek to address it with a horizontally mounted 2D laser scanner. 2D laser scanner is equipped on almost every autonomous vehicle for its superiorities in the field of view, lighting invariance, high accuracy and relatively low price. In this paper, we propose a highly efficient search-based L-Shape fitting algorithm for detecting positions and orientations of vehicles with a 2D laser scanner. Differing from the approach to formulating LShape fitting as a complex optimization problem, our method decomposes the L-Shape fitting into two steps: L-Shape vertexes searching and L-Shape corner localization. Our approach is computationally efficient due to its minimized complexity. In on-road experiments, our approach is capable of adapting to various circumstances with high efficiency and robustness.

cs.RO↗

arXiv2018

Salience Biased Loss for Object Detection in Aerial Images

Object detection in remote sensing, especially in aerial images, remains a challenging problem due to low image resolution, complex backgrounds, and variation of scale and angles of objects in images. In current implementations, multi-scale based and angle-based networks have been proposed and generate promising results with aerial image detection. In this paper, we propose a novel loss function, called Salience Biased Loss (SBL), for deep neural networks, which uses salience information of the input image to achieve improved performance for object detection. Our novel loss function treats training examples differently based on input complexity in order to avoid the over-contribution of easy cases in the training process. In our experiments, RetinaNet was trained with SBL to generate an one-stage detector, SBL-RetinaNet. SBL-RetinaNet is applied to the largest existing public aerial image dataset, DOTA. Experimental results show our proposed loss function with the RetinaNet architecture outperformed other state-of-art object detection models by at least 4.31 mAP, and RetinaNet by 2.26 mAP with the same inference speed of RetinaNet.

cs.CV↗

arXiv2018

Deep Anticipation: Light Weight Intelligent Mobile Sensing in IoT by Recurrent Architecture

The rapid growth of IoT era is shaping the future of mobile services. Advanced communication technology enables a heterogeneous connectivity where mobile devices broadcast information to everything. Mobile applications such as robotics and vehicles connecting to cloud and surroundings transfer the short-range on-board sensor perception system to long-range mobile-sensing perception system. However, the mobile sensing perception brings new challenges for how to efficiently analyze and intelligently interpret the deluge of IoT data in mission- critical services. In this article, we model the challenges as latency, packet loss and measurement noise which severely deteriorate the reliability and quality of IoT data. We integrate the artificial intelligence into IoT to tackle these challenges. We propose a novel architecture that leverages recurrent neural networks (RNN) and Kalman filtering to anticipate motions and interac- tions between objects. The basic idea is to learn environment dynamics by recurrent networks. To improve the robustness of IoT communication, we use the idea of Kalman filtering and deploy a prediction and correction step. In this way, the architecture learns to develop a biased belief between prediction and measurement in the different situation. We demonstrate our approach with synthetic and real-world datasets with noise that mimics the challenges of IoT communications. Our method brings a new level of IoT intelligence. It is also lightweight compared to other state-of-the-art convolutional recurrent architecture and is ideally suitable for the resource-limited mobile applications.

cs.CV↗

arXiv2016

Geography and distance effect on financial dynamics in the Chinese stock market

Geography effect is investigated for the Chinese stock market including the Shanghai and Shenzhen stock markets, based on the daily data of individual stocks. The Shanghai city and the Guangdong province can be identified in the stock geographical sector. By investigating a geographical correlation on a geographical parameter, the stock location is found to have an impact on the financial dynamics, except for the financial crisis time of the Shenzhen market. Stock distance effect is further studied, with a crossover behavior observed for the stock distance distribution. The probability of the short distance is much greater than that of the long distance. The average stock correlation is found to weakly decay with the stock distance for the Shanghai stock market, but stays nearly stable for different stock distance for the Shenzhen stock market.

physics.soc-ph↗

arXiv2015

Hierarchical Latent Semantic Mapping for Automated Topic Generation

Much of information sits in an unprecedented amount of text data. Managing allocation of these large scale text data is an important problem for many areas. Topic modeling performs well in this problem. The traditional generative models (PLSA,LDA) are the state-of-the-art approaches in topic modeling and most recent research on topic generation has been focusing on improving or extending these models. However, results of traditional generative models are sensitive to the number of topics K, which must be specified manually. The problem of generating topics from corpus resembles community detection in networks. Many effective algorithms can automatically detect communities from networks without a manually specified number of the communities. Inspired by these algorithms, in this paper, we propose a novel method named Hierarchical Latent Semantic Mapping (HLSM), which automatically generates topics from corpus. HLSM calculates the association between each pair of words in the latent topic space, then constructs a unipartite network of words with this association and hierarchically generates topics from this network. We apply HLSM to several document collections and the experimental comparisons against several state-of-the-art approaches demonstrate the promising performance.

cs.LG↗

arXiv2013

Heterogeneity Involved Network-based Algorithm Leads to Accurate and Personalized Recommendations

Heterogeneity of both the source and target objects is taken into account in a network-based algorithm for the directional resource transformation between objects. Based on a biased heat conduction recommendation method (BHC) which considers the heterogeneity of the target object, we propose a heterogeneous heat conduction algorithm (HHC), by further taking the source object degree as the weight of diffusion. Tested on three real datasets, the Netflix, RYM and MovieLens, the HHC algorithm is found to present a better recommendation in both the accuracy and personalization than two excellent algorithms, i.e., the original BHC and a hybrid algorithm of heat conduction and mass diffusion (HHM), while not requiring any other accessorial information or parameter. Moreover, the HHC even elevates the recommendation accuracy on cold objects, referring to the so-called cold start problem, for effectively relieving the recommendation bias on objects with different level of popularity.

physics.soc-ph↗

arXiv2012

Promotional effect on cold start problem and diversity in a data characteristic based recommendation method

Pure methods generally perform excellently in either recommendation accuracy or diversity, whereas hybrid methods generally outperform pure cases in both recommendation accuracy and diversity, but encounter the dilemma of optimal hybridization parameter selection for different recommendation focuses. In this article, based on a user-item bipartite network, we propose a data characteristic based algorithm, by relating the hybridization parameter to the data characteristic. Different from previous hybrid methods, the present algorithm adaptively assign the optimal parameter specifically for each individual items according to the correlation between the algorithm and the item degrees. Compared with a highly accurate pure method, and a hybrid method which is outstanding in both the recommendation accuracy and the diversity, our method shows a remarkably promotional effect on the long-standing challenging problem of the cold start, as well as the recommendation diversity, while simultaneously keeps a high overall recommendation accuracy. Even compared with an improved hybrid method which is highly efficient on the cold start problem, the proposed method not only further improves the recommendation accuracy of the cold items, but also enhances the recommendation diversity. Our work might provide a promising way to better solving the personal recommendation from the perspective of relating algorithms with dataset properties.

cs.IR↗

arXiv2011

Dynamics of Bid-ask Spread Return and Volatility of the Chinese Stock Market

Bid-ask spread is taken as an important measure of the financial market liquidity. In this article, we study the dynamics of the spread return and the spread volatility of four liquid stocks in the Chinese stock market, including the memory effect and the multifractal nature. By investigating the autocorrelation function and the Detrended Fluctuation Analysis (DFA), we find that the spread return is lack of long-range memory, while the spread volatility is long-range time correlated. Moreover, by applying the Multifractal Detrended Fluctuation Analysis (MF-DFA), the spread return is observed to possess a strong multifractality, which is similar to the dynamics of a variety of financial quantities. Differently from the spread return, the spread volatility exhibits a weak multifractal nature.

q-fin.ST↗

arXiv2010

Cooperation in the snowdrift game on directed small-world networks under self-questioning and noisy conditions

Cooperation in the evolutionary snowdrift game with a self-questioning updating mechanism is studied on annealed and quenched small-world networks with directed couplings. Around the payoff parameter value $r=0.5$, we find a size-invariant symmetrical cooperation effect. While generally suppressing cooperation for $r>0.5$ payoffs, rewired networks facilitated cooperative behavior for $r<0.5$. Fair amounts of noise were found to break the observed symmetry and further weaken cooperation at relatively large values of $r$. However, in the absence of noise, the self-questioning mechanism recovers symmetrical behavior and elevates altruism even under large-reward conditions. Our results suggest that an updating mechanism of this type is necessary to stabilize cooperation in a spatially structured environment which is otherwise detrimental to cooperative behavior, especially at high cost-to-benefit ratios. Additionally, we employ component and local stability analyses to better understand the nature of the manifested dynamics.

physics.soc-ph↗

arXiv2010

Memory effect and multifractality of cross-correlations in financial markets

An average instantaneous cross-correlation function is introduced to quantify the interaction of the financial market of a specific time. Based on the daily data of the American and Chinese stock markets, memory effect of the average instantaneous cross-correlations is investigated over different price return time intervals. Long-range time-correlations are revealed, and are found to persist up to a month-order magnitude of the price return time interval. Multifractal nature is investigated by a multifractal detrended fluctuation analysis.

q-fin.ST↗

arXiv2010

Adaptive financial networks with static and dynamic thresholds

Based on the daily data of American and Chinese stock markets, the dynamic behavior of a financial network with static and dynamic thresholds is investigated. Compared with the static threshold, the dynamic threshold suppresses the large fluctuation induced by the cross-correlation of individual stock prices, and leads to a stable topological structure in the dynamic evolution. Long-range time-correlations are revealed for the average clustering coefficient, average degree and cross-correlation of degrees. The dynamic network shows a two-peak behavior in the degree distribution.

q-fin.ST↗