Learning about artificial intelligence (AI) has become one of the most discussed topics in field education. However, it an equally important learning approach contemporary education to propose a “general education” agenda that conveys instructional messages AI basics and ethics, especially for those students without engineering background. The current study proposes situated design on this topic. Through three-week lesson session accompanying activities, participants undertook hands-on tasks...

Conventional deep neural networks face challenges in handling the increasing amount of information real-world scenarios where it is impractical to gather all training data at once. Incremental learning, also known as continual provides a solution for lightweight and sustainable learning with networks. However, incremental encounters issues such “catastrophic forgetting” “stability–plasticity dilemma”. To address these challenges, this study proposes two-stage method. In first stage, dual...

In recent years, generative AI has become popular in areas like natural language processing, as well image and audio significantly expanding AI’s creative capabilities. Particularly the realm of generation, diffusion models have achieved remarkable success across various applications, such synthesis transformation. However, traditional operate at pixel level when learning features, which inevitably demands significant computational resources. To address this issue, paper proposes a pose...

Pose transfer methods often struggle to simultaneously preserve fine-grained clothing textures and facial details, especially under large pose variations. To address these limitations, we propose a model based on the Multi-scale attention guided model, with modifications its residual block by integrating convolutional module changing activation function from ReLU Mish capture more features related skin color. Additionally, as generated images had differing original image, two different...

This study presents a hierarchical lane detection system with the ability to deal both structured and unstructured roads. The proposed classifies environment first before applying suitable algorithms for different types of Instead dealing all situations one complicated algorithm, this architecture makes it possible achieve high accuracy relatively simple efficient algorithms. For classification, pixels lane-marking colours are extracted as feature points. Eigenvalue decomposition regularised...

Abstract. This work performs cloud classification on all-sky images. To deal with mixed types in one image, we propose performing block division and block-based classification. In addition to classical statistical texture features, the proposed method incorporates local binary pattern, which extracts features feature vector. The combined can effectively preserve global information as well more discriminating of different types. experimental results have shown that applying higher accuracy...

In this work, an accurate solar irradiance now-casting system is presented. The proposed utilizes information from all-sky images to complement insufficient temporal and spatial resolution of satellite help improve prediction accuracy. Based on the history image features, are predicted ten fifteen minutes ahead allow Photovoltaics operators schedule allocate energy resources more effectively. To capture relationship between clouds sun, enhanced cloud detection algorithm using multi-scale...

This paper presents a system applied to unmanned aerial vehicles based on Robot Operating Systems (ROSs). The study addresses the challenges of efficient object detection and real-time target tracking for vehicles. utilizes pruned YOLOv4 architecture fast SiamMask model continuous tracking. A Proportional Integral Derivative (PID) module adjusts flight attitude, enabling stable automatically in indoor outdoor environments. contributions this work include exploring feasibility pruning...

Photovoltaics modules are usually installed with a tilt angle to improve performance and avoid water or dust accumulation. However, measured irradiance data on inclined surfaces rarely available, since installing pyranometers various angles induces high costs. Estimating of arbitrary orientations is important because the installation might be different at sites. The goal this work propose unified transfer model obtain solar orientations. Artificial neural networks (ANN) were utilized...

In this paper, we propose an intelligent highway surveillance system that performs self-diagnosis and detects conditions when the camera is seriously tampered by raindrops at night. The also provides solutions to analyze traffic flow under challenging nighttime raindrop-tampered conditions. To deal with scenes, extract effective features via salient region detection block segmentation. extracted are used train a support vector machine achieve self-diagnosis. For analysis, use in of interest...

Extracting the flight trajectory of shuttlecock in a single turn badminton games is important for automated sports analytics. This study proposes novel method to extract shots from monocular camera. First, TrackNet, deep neural network designed tracking small objects, used shuttlecock. Second, YOLOv7 model identify whether player swinging. As both TrackNet and may have detection misses false detections, this shot refinement algorithm obtain correct hitting moment. By doing so, we can rallies...

Advancements in satellite and aerial imagery technology have made it easier to obtain high-resolution remote sensing images, leading widespread research applications various fields. Remote image semantic segmentation is a crucial task that provides localization information for target objects. In addition the large-scale variation issues common most datasets, images present unique challenges, including high background complexity imbalanced foreground-background ratios. However, general...

Average Motion Energy (AME) image is a good way to describe human motions. However, it has face the computation efficiency problem with increasing number of database templates. In this paper, we propose histogram-based approach improve efficiency. We convert action/gait recognition histogram matching problem. order speed up process, adopt multiresolution structure on Histogram (MEH). To utilize more efficiently, an automated uneven partitioning method which achieved by utilizing quadtree...

This paper proposes a high-performance advanced driver assistance system that analyses driving scenes based on monocular cameras. The identifies the ego-lane position and indicates if vehicle is an inner or outer lane. Dense optical flow analysis performed fuzzy designed to achieve identification. An event warning application implemented hierarchical classifier results of proposed accurately issues events without having detect vehicles first, making more responsive potential approaching...

While the use of deep neural networks is popular for predicting students’ learning outcomes, convolutional network (CNN)-based methods are used more often. Such require numerous features, training data, or multiple models to achieve week-by-week predictions. However, many current management systems (LMSs) operated by colleges cannot provide adequate information. To make system feasible, this article proposes a recurrent (RNN)-based framework identify at-risk students who might fail course...

In this paper we developed a system for automated human body tracking and modeling based on monocular camera. system, eleven joint points including head, shoulder, hip, elbows, knees, hands feet are extracted separately to build 2D model. The head is by analyzing negative minimum curvature (NMC) parameterized silhouette. torso, along with its angle size, determined integrating multiple frame information connectivity constraint. Hands can be identified correctly modified star skeleton...

This paper presents a hierarchical lane detection system with the ability to deal both structured and unstructured roads. The proposed classifies environment first before applying suitable algorithms for different types of For classification, pixels lane- marking colors are extracted as feature points. Eigenvalue decomposition regularized discriminant analysis is utilized in model selection maximum likelihood estimation Gaussian parameters high dimensional space. roads, points reused...

This article proposes a video-based identification authentication framework via signature verification and pen-grasping posture analysis. The authors consider the case of using camera instead pressure-sensitive tablet to acquire signatures. proposed reliable method is useful when digitising tablets are not available. In addition, can more information in addition trajectories handwritten verification. entire writing process personalised features that cannot be easily imitated forged. analyse...

To most learners, programming language is not simple. In particular, when a large number of errors occur, teaching progress courses will be affected, which leads to poor learning effectiveness. This study developed web-based platform solve this issue.