Junseok Kwon's research while affiliated with Chung-Ang University and other places

Publications (74)

Article
In this paper, we present a novel 3D point cloud harvesting method, which can harvest 3D points from an estimated surface distribution in an unsupervised manner (i.e., an input is a prior distribution). Our method outputs the surface distribution of a 3D object and samples 3D points from the distribution based on the proposed progressive random sam...
Article
Full-text available
Style transfer has attracted a lot of attentions, as it can change a given image into one with splendid artistic styles while preserving the image structure. However, conventional approaches easily lose image details and tend to produce unpleasant artifacts during style transfer. In this paper, to solve these problems, a novel artistic stylization...
Article
Full-text available
Playing multiple stage videos of a particular singer as if they are one is called Stagemix video. The consumption of video media has increased recently, and the demand for video editing has also increased. Stagemix videos have gained popularity in various communities, and a number of YouTubers who upload videos with cross-cuts are appearing. In thi...
Article
This study presents a novel Riemannian submanifold (RS) framework for log-Euclidean metric learning on symmetric positive definite manifolds. Our method identifies the optimal RS without changing the original tangent space. The RS is spanned by multiple bases, and each data point is parameterized using these bases, such that the data can be represe...
Article
Full-text available
We present a novel noise-injected Markov chain Monte Carlo (NMCMC) method for visual tracking, which enables fast convergence through adversarial attacks. The proposed NMCMC consists of three steps: noise-injected proposal, acceptance, and validation. We intentionally inject noise into the proposal function to cause a shift in a direction that is o...
Preprint
Full-text available
Using convolutional neural networks for 360images can induce sub-optimal performance due to distortions entailed by a planar projection. The distortion gets deteriorated when a rotation is applied to the 360image. Thus, many researches based on convolutions attempt to reduce the distortions to learn accurate representation. In contrast, we leverage...
Article
Full-text available
We formulate the visual tracking problem as a semi-supervised continual learning problem, where only an initial frame is labeled. In contrast to conventional meta-learning based approaches that regard visual tracking as an instance detection problem with a focus on finding good weights for model initialization, we consider both initialization and o...
Article
Full-text available
Owing to the improved representation ability, recent deep learning-based methods enable to estimate scene depths accurately. However, these methods still have difficulty in estimating consistent scene depths under real-world environments containing severe illumination changes, occlusions, and texture-less regions. To solve this problem, in this pap...
Article
Full-text available
In this study, we propose a novel Wasserstein distributional tracking method that can balance approximation with accuracy in terms of Monte Carlo estimation. To achieve this goal, we present three different systems: sliced Wasserstein-based (SWT), projected Wasserstein-based (PWT), and orthogonal coupled Wasserstein-based (OCWT) visual tracking sys...
Article
Full-text available
Learning deep neural networks from noisy labels is challenging, because high-capacity networks attempt to describe data even with noisy class labels. In this study, we propose a self-augmentation method without additional parameters, which handles noisy labeled data based on small-loss criteria. To this end, we use small-loss samples by introducing...
Preprint
Full-text available
Style transfer has attracted a lot of attentions, as it can change a given image into one with splendid artistic styles while preserving the image structure. However, conventional approaches easily lose image details and tend to produce unpleasant artifacts during style transfer. In this paper, to solve these problems, a novel artistic stylization...
Article
We present a novel dehazing framework for real-world images that contain both hazy and low-light areas. Dehazing and low-light enhancements are unified by using an illumination map that is estimated using a proposed convolutional neural network. The illumination map is then used as a component for three different tasks: atmospheric light estimation...
Article
Full-text available
Person re-identification (re-id) aims to identity the same person over multiple cameras; it has been successfully applied to various computer vision applications as a fundamental method. Owing to the development of deep learning, person re-id methods, which typically use triplet networks based on triplet loss, have demonstrated great success. Howev...
Article
Full-text available
Abstract The authors present a novel tracking algorithm based on a factorial hidden Markov model (FHMM) that can utilise the structured information of a target. An FHMM consists of multiple hidden Markov models (HMMs), wherein each HMM aims to represent a different part of the target. Then, the geometric relation between patches is encoded in the F...
Article
Full-text available
Recently, surveillance systems have been widely used to analyze video recordings captured by surveillance cameras and to detect abnormal or irregular events in real-world scenes. In this study, we present a novel system that detects abnormal events. Unlike conventional methods, we consider abnormal event detection as variation matching problems. In...
Article
We propose a highly generative dehazing method based on pixel-wise Wasserstein autoencoders. In contrast to existing dehazing methods based on generative adversarial networks, our method can produce a variety of dehazed images with different styles. It significantly improves the dehazing accuracy via pixel-wise matching from hazy to dehazed images...
Article
In this paper, we propose a novel visual tracking method based on conditional uncertainty minimization (CUM), minibatch Monte Carlo (MMC), and non-nested sampling (NNS). We represent a target as a Markov network with nodes and edges, where each node corresponds to the corresponding pixel of the target and each edge describes the relations among the...
Chapter
Recent advances in Siamese network-based visual tracking methods have enabled high performance on numerous tracking benchmarks. However, extensive scale variations of the target object and distractor objects with similar categories have consistently posed challenges in visual tracking. To address these persisting issues, we propose novel TridentAli...
Article
Full-text available
We propose a novel generative adversarial network for class-conditional data augmentation (i.e., GANDA) to mitigate data imbalance problems in image classification tasks. The proposed GANDA generates minority class data by exploiting majority class information to enhance the classification accuracy of minority classes. For stable GAN training, we i...
Article
Full-text available
In this study, we present a novel tracking system, in which the tracking accuracy can be considerably enhanced by state prediction. Accordingly, we present a new Q-learning-based reinforcement method, augmented by Wang–Landau sampling. In the proposed method, reinforcement learning is used to predict a target configuration for the subsequent frame,...
Preprint
In this paper, we present a novel low-light image enhancement method called dark region-aware low-light image enhancement (DALE), where dark regions are accurately recognized by the proposed visual attention module and their brightness are intensively enhanced. Our method can estimate the visual attention in an efficient manner using super-pixels w...
Article
We propose a novel integral probability metric-based generative adversarial network (GAN), called SphereGAN. In the proposed scheme, the distance between two probability distributions (i.e., true and fake distributions) is measured on a hypersphere. Given that its hypersphere-based objective function computes the upper bound of the distance as a ha...
Preprint
Full-text available
Recent advances in Siamese network-based visual tracking methods have enabled high performance on numerous tracking benchmarks. However, extensive scale variations of the target object and distractor objects with similar categories have consistently posed challenges in visual tracking. To address these persisting issues, we propose novel TridentAli...
Article
Full-text available
We propose a new rare-event detection method based on quasi-Wang–Landau Monte Carlo (QWLMC) sampling with approximate Bayesian computation (ABC) called QWLMC-ABC. QWLMC-ABC integrates ABC and a Halton sequence into Wang–Landau Monte Carlo (WLMC) sampling methods. The Halton sequence provides an improved proposal function and increases the accuracy...
Article
In this study, we present a novel visual tracker based on the variational auto-encoding Markov chain Monte Carlo (VAE-MCMC) method. A target is tracked over time with the help of multiple geometrically related supporters whose motions correlate with those of the target. Good supporters are obtained using variational auto-encoding techniques that me...
Article
Full-text available
We present an adaptive regularization algorithm that can be effectively applied to the optimization problem in deep learning framework. Our regularization algorithm aims to take into account the fitness of data to the current state of model in the determination of regularity to achieve better generalization. The degree of regularization at each ele...
Preprint
The average accuracy is one of major evaluation metrics for classification systems, while the accuracy deviation is another important performance metric used to evaluate various deep neural networks. In this paper, we present a new ensemble-like fast deep neural network, Harmony, that can reduce the accuracy deviation among categories without degra...
Preprint
We present an adaptive regularization algorithm that can be effectively applied to the optimization problem in deep learning framework. Our regularization algorithm aims to take into account the fitness of data to the current state of model in the determination of regularity to achieve better generalization. The degree of regularization at each ele...
Preprint
We present a novel dehazing and low-light enhancement method based on an illumination map that is accurately estimated by a convolutional neural network (CNN). In this paper, the illumination map is used as a component for three different tasks, namely, atmospheric light estimation, transmission map estimation, and low-light enhancement. To train C...
Article
Full-text available
We propose a new segmentation algorithm based on deep learning. To segment ice hockey players, a fully convolutional network (FCN) is adopted and fine-tuned with our augmented training data. The original FCN has difficulty segmenting small-size objects. To solve this problem, our method divides an input image into four overlapping sub-images and ea...
Preprint
In this paper, we propose a novel generative adversarial network (GAN) for 3D point clouds generation, which is called tree-GAN. To achieve state-of-the-art performance for multi-class 3D point cloud generation, a tree-structured graph convolution network (TreeGCN) is introduced as a generator for tree-GAN. Because TreeGCN performs graph convolutio...
Article
A novel tracking method is proposed, which infers a target state and appearance template simultaneously. With this simultaneous inference, the method accurately estimates the target state and robustly updates the target template. The joint inference is performed by using the proposed particle swarm optimization–Markov chain Monte Carlo (PSO–MCMC) s...
Conference Paper
Full-text available
This paper proposes a low-complexity online model adaptation algorithm which dynamically selects an ob- ject detection algorithm among given/implemented algorithms in the system depending on workload-backlog. As well-studied in literature, there exists tradeoff between object detection accuracy and computation time (i.e., delay) because highly accu...
Article
Here, the authors propose a novel tracking algorithm that can automatically modify the initial configuration of a target to improve the tracking accuracy in subsequent frames. To achieve this goal, the authors' method analyses the likelihood landscape (LL) for the image patch described by the initial configuration. A good configuration has a unimod...
Article
Full-text available
We present a novel tracking system that adaptively selects a shape of the posterior over time, where the selection is efficiently performed by the uncertainty calibrated Markov Chain Monte Carlo (UCMCMC) sampler. In conventional trackers, the posterior is typically described by a single prior distribution. On the other hand, our tracker allows the...
Article
One of the major challenges of model-free visual tracking problem has been the difficulty originating from the unpredictable and drastic changes in the appearance of objects we target to track. Existing methods tackle this problem by updating the appearance model on-line in order to adapt to the changes in the appearance. Despite the success of the...
Article
Full-text available
In this paper, we propose a novel on-line visual tracking framework based on Siamese matching network and meta-learner network which runs at real-time speed. Conventional deep convolutional feature based discriminative visual tracking algorithms require continuous re-training of classifiers or correlation filters for solving complex optimization ta...
Article
In this paper, we present a novel tracking system based on edge-based object proposal and data association called object proposal association. Our object proposal method accurately detects and localizes objects in an image by searching for object-like regions, with the assumption that an object is represented by a closed boundary. To search for clo...
Article
Full-text available
One of the major challenges of model-free visual tracking problem has been the difficulty originating from the unpredictable and drastic changes in the appearance of objects we target to track. Existing methods tackle this problem by updating the appearance model on-line in order to adapt to the changes in the appearance. Despite the success of the...
Article
In this paper, the accuracy of visual tracking is enhanced by leveraging a novel measure for observation quality. We measure observation quality with mutual information, then look at the interval covered by that mutual information. As observation uncertainty the interval length is proposed. The best observation is considered the one that both maxim...
Article
Visual tracking is one of the computer vision’s longstanding challenges, with many methods as a result. While most state-of-the-art methods trade-off performance for speed, we propose PICASO, an efficient, yet strongly performing tracking scheme. The target object is modeled as a set of pixel-level templates with weak configuration constraints. The...
Article
We propose a novel tracking method that allows to switch between different state representations as, e.g., image coordinates in different views or image and ground plane coordinates. During the tracking process, our method adaptively switches between these representations. We demonstrate the applicability of our method for dynamic cameras tracking...
Article
We propose a novel framework that jointly estimates the ground plane and a target's motion trajectory. This results in improvements for both. Estimating their joint posterior is based on Particle Markov Chain Monte Carlo (Particle MCMC). In Particle MCMC, the best target state is inferred by a Particle Filter and the best ground plane is obtained b...
Article
A novel tracking algorithm is proposed, which robustly tracks a target by finding the state that minimizes the likelihood uncertainty. Likelihood uncertainty is estimated by determining the gap between the lower and upper bounds of likelihood. By minimizing the gap between the two bounds, the proposed method identifies the confident and reliable st...
Conference Paper
Visual tracking is the task of estimating the trajectory of an object in a video given its initial location. This is usually done by combining at each step an appearance and a motion model. In this work, we learn from a small set of training trajectory annotations how the objects in the scene typically move. We learn the relative weight between the...
Conference Paper
A novel tracking algorithm that can track a highly non-rigid target robustly is proposed using a new bounding box representation called the Double Bounding Box (DBB). In the DBB, a target is described by the combination of the Inner Bounding Box (IBB) and the Outer Bounding Box (OBB). Then our objective of visual tracking is changed to find the IBB...
Article
We propose the visual tracker sampler, a novel tracking algorithm that can work robustly in challenging scenarios, where several kinds of appearance and motion changes of an object can occur simultaneously. The proposed tracking algorithm accurately tracks a target by searching for appropriate trackers in each frame. Since the real-world tracking e...
Conference Paper
This paper proposes a robust tracking method that uses interval analysis. Any single posterior model necessarily includes a modeling uncertainty (error), and thus, the posterior should be represented as an interval of probability. Then, the objective of visual tracking becomes to find the best state that maximizes the posterior and minimizes its in...
Article
A novel tracking algorithm is proposed for targets with drastically changing geometric appearances over time. To track such objects, we develop a local patch-based appearance model and provide an efficient online updating scheme that adaptively changes the topology between patches. In the online update process, the robustness of each patch is deter...
Conference Paper
In this paper, a robust visual tracking method is proposed by casting tracking as an estimation problem of the joint space of non-rigid appearance model and state. Conventional trackers which use templates as the appearance model do not handle ambiguous samples effectively. On the other hand, trackers that use non-rigid appearance models have low d...
Conference Paper
We propose a novel tracking algorithm that robustly tracks the target by finding the state which minimizes uncertainty of the likelihood at current state. The uncertainty of the likelihood is estimated by obtaining the gap between the lower and upper bounds of the likelihood. By minimizing the gap between the two bounds, our method finds the confid...
Article
We propose a novel tracking algorithm based on the Wang-Landau Monte Carlo (WLMC) sampling method for dealing with abrupt motions efficiently. Abrupt motions cause conventional tracking methods to fail because they violate the motion smoothness constraint. To address this problem, we introduce the Wang-Landau sampling method and integrate it into a...
Conference Paper
A novel approach for event summarization and rare event detection is proposed. Unlike conventional methods that deal with event summarization and rare event detection independently, we solve them together by transforming the problems into a graph editing framework. In our approach, a video is represented as a graph, in which each node of the graph...
Conference Paper
Recent studies on visual tracking have shown significant improvement in accuracy by handling the appearance variations of the target object. Whereas most studies present schemes to extract the time-invariant characteristics of the target and adaptively update the appearance model, the present paper concentrates on modeling the probabilistic depende...
Conference Paper
We propose a novel tracking framework called visual tracker sampler that tracks a target robustly by searching for the appropriate trackers in each frame. Since the real-world tracking environment varies severely over time, the trackers should be adapted or newly constructed depending on the current situation. To do this, our method obtains several...
Conference Paper
We propose a novel tracking algorithm that can work robustly in a challenging scenario such that several kinds of appearance and motion changes of an object occur at the same time. Our algorithm is based on a visual tracking decomposition scheme for the efficient design of observation and motion models as well as trackers. In our scheme, the observ...
Conference Paper
We propose a novel approach for synchronizing multiple videos and simultaneously detecting rare events in these videos. Unlike conventional methods which deal with video synchronization and rare event detection separately, we cast these problems into an unified energy minimization framework and present a Cross-Entropy Monte Carlo (CEMC) based metho...
Conference Paper
We propose a novel tracking algorithm for the target of which geometric appearance changes drastically over time. To track it, we present a local patch-based appear- ance model and provide an efficient scheme to evolve the topology between local patches by on-line update. In the process of on-line update, the robustness of each patch in the model i...
Conference Paper
We propose a novel tracking algorithm based on the Wang-Landau Monte Carlo sampling method which efficiently deals with the abrupt motions. Abrupt motions could cause conventional tracking methods to fail since they violate the motion smoothness constraint. To address this problem, we introduce the Wang-Landau algorithm that has been recently propo...

Citations

... cosformer [28] replaces non-decomposable nonlinear softmax operations with linear operations with a decomposable nonlinear reweighting mechanism, which not only achieves comparable or better performance than softmax-attention across a range of tasks, but also has linear space and time complexity. [29] introduces softmax into network pruning. Softmax-attention channel pruning consists of training, pruning, and fine-tuning steps. ...
... Recently, GCNs have been shown to be successful in the machine learning domain for graph structure data representation and learning [5]. Kim et al. [18] introduced a novel person re-id approach based on a triplet-structured GCN in this paper. Where the suggested GCN takes advantage of structured information among triplet samples The features of triplet samples are utilized in the graph network's triplet loss to improve the performance. ...
... Kim et al. [12] proposed a method based on pixel-wise Wasserstein autoencoders. The technique uses generative adversarial networks to produce various dehazed images with different styles. ...
... It is an enabler of automation where image based process control and surveillance of industrial operations can be performed (Heras and Blanke, 2020). Machine vision has several applications in industrial surveillance (Cho and Kwon, 2021), traffic monitoring and control system, automated inspection, driver less cars, vehicle guidance and interpretation of remotely sensed images (Hannan et al., 2009). One common application of camera-enabled surveillance is fixed installation roadside cameras used for smart traffic management with real time analysis (Jacob et al., 2019) that capture image of registration number of over speeding and signal jumping vehicles (Fig. 3). ...
... To validate the effectiveness of our proposed framework, we apply our method to Siamese network-based tracking algorithm TACT [29], which is a two-stage detector-based tracker. We compare our method to other state-of-the-art trackers on test splits of large-scale visual tracking datasets, including LaSOT [18], OxUvA [19], TLP [20], Track-ingNet [16], and GOT-10k [17]. ...
... For the first problem discussed in the next section, the authors often adopt general-purpose datasets (such as [42]) and assume the images are rectified. The nature of the problem allows generating an arbitrary number of annotations by synthesizing rotations, and quantitative results are often obtained using angular distances [23], [43], [44]. The recent survey from [16] compiles tens of datasets and figures of merit (omitted here due to space restrictions) for the other two applications discussed in the next section. ...
... These works do not explore the prospect of imbalanced classification between the domains, and do not utilize the synthesized samples to train a further classifier. Some approaches exist that aim to augment a minority class, but they are mostly focused on visual image tasks [16,17]. Oversampling via translation Translation-based approaches to oversampling are relatively underdeveloped. ...
... [140] -SSGAN [6] -PPL Reg. [8] -LGAN [139] -CRGAN [39] -LOGAN [125] -Top-k Training [141] -ICRGAN [126] -cGAN [75] -ACGAN [76] -Improved GAN [27] -AMGAN [77] -cBN [78,91,92] -PD-GAN [78] -WCBT [79] -TACGAN [80] -MHGAN [81] -ContraGAN [82] -OmniGAN [83] -ADCGAN [84] -ReACGAN [10] -pix2pix [64] -CycleGAN [11] -UNIT [65] -DiscoGAN [66] -StarGAN [13] -MUNIT [67] -InstaGAN [68] -DRIT [69] -FUNIT [70] -U-GAT-IT [71] -StarGAN v2 [72] -CUT [73] -ICGAN [74] -Vanilla GAN [1] -f-GAN [3] -LSGAN [134] -EBGAN [5] -Unrolled GAN [4] -Geometric GAN [135] -WGAN [93] -McGan [98] -MMD GAN [99] -Fisher GAN [100] -Cramér GAN [101] -SphereGAN [102] Conditioning goal -DCGAN [2] -ResNetGAN [55] -Progressive GAN [56] -SAGAN [34] -BigGAN [7] -BigGAN-Deep [7] -StyleGAN [57] -StyleGAN2 [8] -FastGAN [58] -TransGAN [59] -HiT [60] -StyleGAN3 [9] -Projected GAN [61] -ViTGAN [62] -StyleGAN-XL [63] Data-efficient Regularization IPM [97] Non-IPM Class category Image ...
... For example, SP-GAN [22] proposes an unsupervised sphere-guided generative model for directly synthesize point clouds. Others propose various CNN-based GAN architectures [1,17,33] to generate novel shapes in the point cloud format. ...
... Online model updating can adaptively learn target features and mitigate "model drift" issues. There has been a lot of work on this aspect recently, such as MLT [6], GradNet [35] , UpdateNet [59] and LTMU [10], and they are all focused on template updating. Different from online model updating, we are among the first to utilize the comparative response between the target and search branch to update search features from bottom to top via attentions. ...