Pedestrian behavior prediction is one of the most critical tasks in urban driving scenarios, playing a key role in ensuring road safety. Traditional learning-based methods have relied on vision models for pedestrian behavior prediction. However, fully understanding pedestrians’ behaviors in advance is very challenging due to the complex driving environments and the multifaceted interactions between pedestrians and road elements. Additionally, these methods often show a limited understanding of driving environments not included in the training. The emergence of Multimodal Large Language Models (MLLMs) provides an innovative approach to addressing these challenges through advanced reasoning capabilities. This paper presents OmniPredict, the first study to apply GPT-4o(mni), a state-of-the-art MLLM, for pedestrian behavior prediction in urban driving scenarios. We assessed the model using the JAAD and WiDEVIEW datasets, which are widely used for pedestrian behavior analysis. Our method utilized multiple contextual modalities and achieved 67% accuracy in a zero-shot setting without any task-specific training, surpassing the performance of the latest MLLM baselines by 10%. Furthermore, when incorporating additional contextual information, the experimental results demonstrated a significant increase in prediction accuracy across four behavior types (crossing, occlusion, action, and look). We also validated the model s generalization ability by comparing its responses across various road environment scenarios. OmniPredict exhibits strong generalization capabilities, demonstrating robust decision-making in diverse and unseen driving rare scenarios. These findings highlight the potential of MLLMs to enhance pedestrian behavior prediction, paving the way for safer and more informed decision-making in road environments.

This work remedies the drawbacks of existing methods in pedestrian trajectory prediction for video surveillance. A traditional two-stage approach, which first detects and tracks pedestrians before forecasting their subsequent trajectories, relies solely on historical trajectories to predict future movements, overlooking the rich contextual information within videos. Although some one-stage methods have been introduced to streamline this process, they still struggle with complex model architectures due to their reliance on pixel-wise flow estimation. To resolve these issues, we propose a simple yet effective Transformer-based one-stage model, called PredDETR, that directly anticipates the future trajectories of multiple pedestrians from videos. Following the end-to-end paradigm of DETR, our PredDETR model not only associates the bounding boxes of each pedestrian across subsequent frames but also predicts their future trajectories using an additional predictive decoder. Taking past raw video frames as input, the proposed PredDETR directly forecasts the future locations of each pedestrian in a non-autoregressive manner. The empirical results validate that, despite its simpler design, PredDETR is a compelling method compared to a previous one-stage approach.

We address a weakly-supervised low-shot instance segmentation, an annotation-efficient training method to deal with novel classes effectively. Since it is an under-explored problem, we first investigate the difficulty of the problem and identify the performance bottleneck by conducting systematic analyses of model components and individual sub-tasks with a simple baseline model. Based on the analyses, we propose ENInst with sub-task enhancement methods: instance-wise mask refinement for enhancing pixel localization quality and novel classifier composition for improving classification accuracy. Our proposed method lifts the overall performance by enhancing the performance of each sub-task. We demonstrate that our ENInst is 7.5 times more efficient in achieving comparable performance to the existing fully-supervised few-shot models and even outperforms them at times.

In this letter, we present a novel approach for denoising channel state information (CSI) feedback in massive multiple-input multiple-output (MIMO) cellular networks. Our method utilizes Deep Learning (DL) techniques to compress and remove noise from measured CSI. Traditional DL-based denoising requires pairs of noisy input and corresponding clean targets, which are impractical to obtain in real-world wireless networks. To address this challenge, we propose a training method of denoising autoencoder using pairs of noisy CSIs and practical data acquisition strategies. Extensive evaluations demonstrate the superior reconstruction performance of our method compared to a vanilla autoencoder and legacy codebook-based CSI feedback.

As the importance of trajectory analysis arises in video surveillance, it becomes crucial to define the dissimilarity measure between two trajectories. Although the Hausdorff distance can be considered as a viable candidate for the measure, it is challenging to deal with noise present in trajectories since the Hausdorff distance is susceptible to noise so that even a single noise point may significantly distort the distance computation. In this paper, we propose a novel approach to alleviate the influence of inherent noise by setting noise-like points apart from ordinary points with a novel spatial tree structure during trajectory distance computation, without additional noise detection processes. In particular, we present R-on -tree, an extension of the existing spatial tree structure, that seamlessly finds permanent noise-like points, which are considered to have a low possibility of being ordinary points, and then keeps them in a separate auxiliary R-tree, without any separate process of disclosing noise-like points. We exploit R-on-tree to compute the noise-tolerant trajectory distance by modifying an existing algorithm for the Hausdorff distance. We also build an algorithm for noise-tolerant trajectory search to ensure accurate and high-quality search results even with noisy trajectories. The empirical results show that in all cases, our proposed approach yields the distance closest to the true one than any other competitor. The effectiveness of our approach is further examined by applying our noise-tolerant trajectory search to a real video surveillance dataset.

We present Uni-DVPS, a unified model for Depth-aware Video Panoptic Segmentation (DVPS) that jointly tackles distinct vision tasks, i.e., video panoptic segmentation, monocular depth estimation, and object tracking. In contrast to the prior works that adopt diverged decoder networks tailored for each task, we propose an architecture with a unified Transformer decoder network. We design a single Transformer decoder network for multi-task learning to increase shared operations to facilitate the synergies between tasks and exhibit high efficiency. We also observe that our unified query learns instance-aware representation guided by multi-task supervision, which encourages query-based tracking and obviates the need for training extra tracking module. We validate our architectural design choices with experiments on Cityscapes-DVPS and SemKITTI-DVPS datasets. The performances of all tasks are jointly improved, and we achieve state-of-the-art results on DVPQ metric for both datasets.

In this paper, we propose a bidirectional style transfer method by exchanging the style of inputs while preserving the structural information. The proposed bidirectional style transfer network consists of three modules: 1) content and style extraction module that extracts the structure and style-related features, 2) local feature transform module that aligns locally extracted feature to its original coordinate, and 3) reconstruction module that generates a newly stylized image. Given two input images, we extract content and style information from both images in a global and local manner, respectively. Note that the content extraction module removes style-related information by compressing the dimension of the feature tensor to a single channel. The style extraction module removes content information by gradually reducing the spatial size of a feature tensor. The local feature transform module exchanges the style information and spatially transforms the local features to its original location. By substituting the style information with one another in both ways (i.e., global and local) bidirectionally, the reconstruction module generates a newly stylized image without diminishing the core structure. Furthermore, we enable the proposed network to control the degree of style to be applied when exchanging the style of inputs bidirectionally. Through the experiments, we compare the bidirectionally style transferred results with existing methods quantitatively and qualitatively. We show generation results by controlling the degree of applied style and adopting various textures to an identical structure.

Since many safety-critical systems, such as surgical robots and autonomous driving cars operate in unstable environments with sensor noise and incomplete data, it is desirable for object detectors to take the localization uncertainty into account. However, there are several limitations of the existing uncertainty estimation methods for anchor-based object detection. 1) They model the uncertainty of the heterogeneous object properties with different characteristics and scales, such as location (center point) and scale (width, height), which could be difficult to estimate. 2) They model box offsets as Gaussian distributions, which is not compatible with the ground truth bounding boxes that follow the Dirac delta distribution. 3) Since anchor-based methods are sensitive to anchor hyper-parameters, their localization uncertainty could also be highly sensitive to the choice of hyper-parameters. To tackle these limitations, we propose a new localization uncertainty estimation method called UAD for anchor-free object detection. Our method captures the uncertainty in four directions of box offsets (left, right, top, bottom) that are homogeneous, so that it can tell which direction is uncertain, and provide a quantitative value of uncertainty in [0, 1]. To enable such uncertainty estimation, we design a new uncertainty loss, negative power log-likelihood loss, to measure the localization uncertainty by weighting the likelihood loss by its IoU, which alleviates the model misspecification problem. Furthermore, we propose an uncertainty-aware focal loss for reflecting the estimated uncertainty to the classification score. Experimental results on COCO datasets demonstrate that our method significantly improves FCOS, by up to 1.8 points, without sacrificing computational efficiency.

In this paper, we propose the duality-based image sequence matching method, which is called Dual-ISM, a subsequence matching method for searching for similar images. We first extract feature points from the given image data and configure the feature vectors as one data sequence. Next, the feature vectors are configured in the form of a disjoint window, and a low-dimensional transformation is carried out. Subsequently, the query image that is entered to construct the candidate set is similarly subjected to a low-dimensional transformation, and the low-dimensional transformed window of the data sequence and window that are less than the allowable value, ε, is regarded as the candidate set using a distance calculation. Finally, similar images are searched in the candidate set using the distance calculation that are based on the original feature vector.

In this paper, we present a two-stage scalable channel estimator (TSCE), a deep learning (DL)-based scalable, and robust channel estimator for wireless cellular networks, which is made up of two DL networks to efficiently support different resource allocation sizes and reference signal configurations. Both networks use the transformer, one of cutting-edge neural network architecture, as a backbone for accurate estimation. For computation-efficient global feature extractions, we propose using window and window averaging-based self-attentions. Our results show that TSCE learns wireless propagation channels correctly and outperforms both traditional estimators and baseline DL-based estimators. Additionally, scalability and robustness evaluations are performed, revealing that TSCE is more robust in various environments than the baseline DL-based estimators.

The effectiveness of rear-flow-controlled mist chemical vapor deposition (mist CVD) for the growth of 2-inch α -Ga 2 O 3 epilayers was studied. The numerical simulation indicated that the low velocity of the flow is appropriate for inducing an upward flow on the growth front without a vortex. Under the flow velocity of 0.08 m/s, α -Ga 2 O 3 the epilayers were successfully grown on c -plane sapphire substrates. The epilayers were high-quality with full widths at half maximum of 42 arcsec and 1993 arcsec for the (0006) and (104) plane reflections, respectively. The rear-flow-controlled mist CVD was demonstrated to be effective for long-time growth. The thickness was adequately increased with increasing growth time. At the same time, corundum α -phase crystal features were distinguished. The suggested mist CVD system not only provides a cost-saving solution for Ga 2 O 3 epilayers’ growth but is also effective for retaining the uniformity of the Ga 2 O 3 epilayers over a large area.

Recently, the need for exoskeleton robots has been increased due to the advancement of robotic technologies and changes in the concept of how the robots can be utilized in direct contact with human bodies. The robots, once only used on the factory floor, are now becoming a part of human bodies, which provides the unprecedented level of muscle power boost and the increase of running speed. If used very carefully, the exoskeleton robots can be also used for patients’ rehabilitation. The exoskeleton robots have many potential application areas; hence most advanced countries are currently developing various types of exoskeleton robots. Those robots can be classified into two major categories, namely the rigid type and the soft type. Each type has own advantages and disadvantages, while the carrying load capacity and the actuation speed can be quite different. There are also many technical difficulties in order to use the exoskeleton robots in the field. The aim of this study is, therefore, to introduce the trends of exoskeleton robot development in advanced countries, while providing the analysis on the technical merits and downside of robot types. The comparison chart also indicates the major technical directions, in which the future technology will be headed for, such as the improved robot response characteristics by employing advanced sensors and artificial intelligence. The robots are becoming smarter, lighter, and more powerful. It is foreseeable that the wearable robots can be a part of human life in the very near future.

In order to deliver medical products (medicines, vaccines, blood packs, etc.) in time for needed areas, a method of transporting goods using drones is being studied.However, temperature-sensitive medical products may decay due to outside temperature changes.The time required to transport over the distance may vary a lot as well.As a result, the likelihood of the goods deteriorating is very high.There is a need for a study on cargo bay to prevent this and to protect the medical goods.In this paper, in order to protect the temperature sensitive medical goods, the inside cargo bay is equipped with the cooling fan device and the electric heating elements.These elements can be monitored and controlled according to the user's discretion.By using the web server built inside the cloud server, the temperature can be controlled in real-time from anywhere without the limitation of distance.We built the proposed device, and installed it on the drone cargo bay.The test results show that the cargo bay can be temperature-controlled, and the setting can be maintained over a great distance.The user can watch the temperature variations during the transport and ascertain the goodness of the medical supply with the data.It is expected that such development can greatly enhance the utility of the drone operations, especially for the medical supply transport applications.

alpha-Ga2O3 thin films were grown on a c-plane sapphire substrate by mist chemical vapor deposition in a horizontal furnace. The microstructure of the alpha-Ga2O3 grown layers was confirmed by X-ray diffraction. The effects of the temperature distribution and velocity field of the mist flow at the surface of the substrate on the uniformity of the epitaxy layers were revealed by numerical simulations and experimental measurements. The thickness uniformity of the thin film was found to be dependent on the position of the inlet and on the flow rate. The suggested configuration for obtaining highly uniform Ga2O3 epilayers is a low flow rate of less than 0.1 m/s and the flow directed from the bottom of the substrate to the top surface of the substrate. (c) The Author(s) 2019. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited.

In this paper, we propose a new framework for detecting the unauthorized dumping of garbage in real‐world surveillance camera. Although several action/behavior recognition methods have been investigated, these studies are hardly applicable to real‐world scenarios because they are mainly focused on well‐refined datasets. Because the dumping actions in the real‐world take a variety of forms, building a new method to disclose the actions instead of exploiting previous approaches is a better strategy. We detected the dumping action by the change in relation between a person and the object being held by them. To find the person‐held object of indefinite form, we used a background subtraction algorithm and human joint estimation. The person‐held object was then tracked and the relation model between the joints and objects was built. Finally, the dumping action was detected through the voting‐based decision module. In the experiments, we show the effectiveness of the proposed method by testing on real‐world videos containing various dumping actions. In addition, the proposed framework is implemented in a real‐time monitoring system through a fast online algorithm.

For contingencies occurring in a complex marine environment, Unmanned Surface Vehicle (USV) has to recognize the situation and decide behaviors, and plan the following actions through the Integrated Mission Planning Process. Therefore, researches are actively being carried out about it. However, since it is difficult to test the actual USV with the mission planning process, it is necessary to develop a virtual experimental environment based on Modeling & Simulation (M&S). In this study, we developed an integrated simulation environment capable of simulating and analyzing the overall mission of USV. In Part I, we modelled the USV Integrated Mission Planning Process and in Part II, we developed an experimental framework and interface for loading them. In addition, we verified the suitability of this model through scenarios and defined the Mission of Effectiveness (MOE) concept for USV mission analysis.

Analyses of social media have increased in importance for understanding human behaviors, interests, and opinions. Business intelligence based on social media can reduce the costs of managing customer trend complexities. This paper focuses on analyzing sensation information representing human perceptual experiences in social media through the five senses: sight, hearing, touch, smell, and taste. First a measurement is defined to estimate social sensation intensities, and subsequently sensation characteristics on geo-social media are identified using geo-spatial footprints. Finally, we evaluate the accuracy and F-measure of our approach by comparing with baselines.

V-doped semi-insulating (VDSI) SiC crystal is a promising substrate for high-frequency electronic devices achieved using GaN epitaxial films. However, V doping in a SiC crystal is difficult to control owing to the different sublimation temperatures of VC and SiC. The amount of V changes depending on the growth sequence, which has been a significant concern in VDSI SiC substrates in terms of wafer reliability.In this study, therefore, we aimed to synthesize a single source by vaporizing Si, C, and V under the same conditions to improve the doping issue in VDSI SiC. We synthesized V-doped SiC powder as the starting material for VDSI SiC substrate based on thermodynamic modeling, and the synthesized powder was used to grow a VDSI SiC crystal via physical vapor transport.Finally, considering the homogeneous V concentration in the grown crystal, the synthesized V-doped SiC was observed to be effective to grow VDSI SiC independent of the growth sequence.