Phet Aimtongkham

Tomato is one of Thailand’s most significant economic crops, generating substantial export value and serving as a primary source of income for local farmers. However, the traditional manual grading process often fails to comply with the Thai Agricultural Standard TACFS 1503–2007, as grading decisions rely heavily on individual experience and subjective perception, resulting in inconsistent quality. Existing automated systems face the challenges of low accuracy, high costs, and complex hardware, while many are incompatible with Thailand’s grading standards. This study presents a multi-view tomato grading system (MVT-Grader), utilizing a dataset acquired from Doi Kham Food Products Co., Ltd. (Third Royal Factory, Tao Ngoi) under controlled lighting conditions. Subsequently, MVT-Grader is built on a custom-designed lightweight CNN architecture with an adjusted spatially aware loss function to enhance the model’s sensitivity in detecting subtle surface defects and color variations. The proposed model was trained using tomato images captured from two and three different viewpoints via a low-cost webcam setup and processed by a GPU-embedded system. Experiments conducted using stratified 5-fold cross-validation on a real-world industrial dataset demonstrate average grading accuracies of 99.43% (two-view) and 99.64% (three-view). Furthermore, the proposed Real-Time Lightweight CNN with Spatially Aware Loss Optimization achieves processing speeds of 87 ms and 114 ms per tomato for two- and three-view cases, respectively. Compared with MVCNN-Siamese, SDF-ConvNets, and Multi-View Spatial Network, the proposed system outperforms the others in both accuracy and speed, improving accuracy by 1.6–6.11% and reducing processing time by 39–49 ms.

Unmanned aerial vehicle (UAV) object detection remains challenging because of the substantial scale variations, frequent small objects, and complex backgrounds present in aerial imagery. In this paper, LECF, a lightweight edge-aware and context-gated feature fusion network designed for multiscale UAV detection, is presented. The network integrates three modules: (1) a multiscale edge feature fusion-enhanced module (MEFFEM) that strengthens edge cues and nonlinear feature representations; (2) a lightweight efficient upsampling-based multibranch feature pyramid network (LEMFPN) that is equipped with an efficient upsampling and spatial shifting (EUSS) component to enhance the cross-scale semantic aggregation process; and (3) a selective gate-based convolution enhancement module (SGCE) that refines feature responses by suppressing redundant activations. When evaluated on VisDrone and SeaDronesSee, compared with YOLOv11s, the AP and AP50 of the LECF model are improved by 2.5% and 3.4% on VisDrone, and by 2.4% and 2.0% on SeaDronesSee, respectively, while the number of required parameters is reduced by 28.7%. The LECF model also provides consistent improvements across multiscale categories and maintains comparable computational complexity, offering a practical lightweight alternative for use in UAV scenarios.

In light of the flourishing proliferation of internet services, the popularity of the Internet of Things (IoT) has swiftly grown in the medical and healthcare fields, and this has been accompanied by a simultaneous escalation in the sophistication of intrusion attacks. Drawing inspiration from the accomplishments of deep learning in cyber threat detection, we propose a multigrained scanning-based deep stacking network (MGDSN) to defend against sophisticated cyberattacks on Internet of Medical Things (IoMT) networks. To address the obscured characteristics of intricate cyberattacks, the MGDSN incorporates four components. First, the feature augmentation process leverages an improved multigrained scanning technique to enhance discriminative information. Second, a deep stacking network (DSN) with a weighting mechanism is employed to generate a set of predictive results for making the final decision. Third, a meta-classifier is introduced to scrutinize the influence of the predictive results when producing the final decision and exploiting a set of meaningfully extracted features. Finally, a loss function is properly designed to take both the predictive losses of the DSN modules and the final predictive loss into account. The outstanding performance achieved by the MGDSN is confirmed through comprehensive evaluations comparing it with the state-of-the-art techniques, encompassing metrics such as the accuracy, precision, recall, F1 score, Cohen’s kappa coefficient, Matthews correlation coefficient, and area under the curve achieved on the IoMT datasets. The MGDSN exhibits a notable improvement ranging from approximately 0.12%-329.21%.

The Internet of Things (IoT) has gained popularity and is widely used in modern society. The growth in the sizes of IoT networks with more internet-connected devices has led to concerns regarding privacy and security. In particular, related to the routing protocol for low-power and lossy networks (RPL), which lacks robust security functions, many IoT devices in RPL networks are resource-constrained, with limited computing power, bandwidth, memory, and battery life. This causes them to face various vulnerabilities and potential attacks, such as DIO neighbor suppression attacks. This type of attack specifically targets neighboring nodes through DIO messages and poses a significant security threat to RPL-based IoT networks. Recent studies have proposed methods for detecting and mitigating this attack; however, they produce high false-positive and false-negative rates in detection tasks and cannot fully protect RPL networks against this attack type. In this paper, we propose a novel fuzzy logic-based intrusion detection scheme to secure the RPL protocol (FLSec-RPL) to protect against this attack. Our method is built of three key phases consecutively: (1) it tracks attack activity variables to determine potential malicious behaviors; (2) it performs fuzzy logic-based intrusion detection to identify malicious neighbor nodes; and (3) it provides a detection validation and blocking mechanism to ensure that both malicious and suspected malicious nodes are accurately detected and blocked. To evaluate the effectiveness of our method, we conduct comprehensive experiments across diverse scenarios, including Static-RPL and Mobile-RPL networks. We compare the performance of our proposed method with that of the state-of-the-art methods. The results demonstrate that our method outperforms existing methods in terms of the detection accuracy, F1 score, power consumption, end-to-end delay, and packet delivery ratio metrics.

Multiuser gigantic-multiple-input multiple-output (MU-gMIMO) and nonorthogonal multiple access (NOMA) are jointly seen as important enabling technologies for sixth generation (6G) networks. They have many benefits, such as spatial multiplexing, spatial diversity, massive connectivity, and spectral efficiency (SE). However, gMIMO-NOMA suffers from many inherent challenges. In this paper, we propose an MU-gMIMO-hybrid multiple-access (HMA) heterogeneous network architecture to address the ‘nearly same channel gain’ issue. Then, an iterative minimal mean squared error (IMMSE) scheme is applied along with quadrature amplitude modulation (QAM) for maximal ratio transmission in the proposed transceiver design to address the ‘residual error’ caused by imperfect successive interference cancellation (ISIC). Finally, to assess the performance of the proposed architecture for the ‘time offsets’ issue, we investigate the asynchronous mode with an MMSE detector matrix following imperfect channel state information (ICSI) to provide a new analysis for HMA transmission and formulate an optimization problem for energy efficiency (EE).

Owing to the exponential proliferation of internet services and the sophistication of intrusions, traditional intrusion detection algorithms are unable to handle complex invasions due to their limited representation capabilities and the unbalanced nature of Internet of Things (IoT)-related data in terms of both telemetry and network traffic. Drawing inspiration from deep learning achievements in feature extraction and representation learning, in this study, we propose an accurate hybrid ensemble deep learning framework (HEDLF) to protect against obfuscated cyber-attacks on IoT networks . To address complex features and alleviate the imbalance problem, the proposed HEDLF includes three key components: (1) a hierarchical feature representation technique based on deep learning, which aims to extract specific information by supervising the loss of gradient information; (2) a balanced rotated feature extractor that simultaneously encourages the individual accuracy and diversity of the ensemble classifier; and (3) a meta-classifier acting as an aggregation method, which leverages a semisparse group regularizer to analyze the base classifiers' outputs. Additionally, these improvements take class imbalance into account. The experimental results show that when compared against state-of-the-art techniques in terms of accuracy, precision, recall, and F1-score, the proposed HEDLF can achieve promising results on both telemetry and network traffic data.

Network congestion is a key challenge in resource‐constrained networks, particularly those with limited bandwidth to accommodate high‐volume data transmission, which causes unfavorable quality of service, including effects such as packet loss and low throughput. This challenge is crucial in wireless sensor networks (WSNs) with restrictions and constraints, including limited computing power, memory, and transmission due to self‐contained batteries, which limit sensor node lifetime. Determining a path to avoid congested routes can prolong the network. Thus, we present a path determination architecture for WSNs that takes congestion into account. The architecture is divided into 3 stages, excluding the final criteria for path determination: (1) initial path construction in a top‐down hierarchical structure, (2) path derivation with energy‐aware assisted routing, and (3) congestion prediction using exponential smoothing. With several factors, such as hop count, remaining energy, buffer occupancy, and forwarding rate, we apply fuzzy logic systems to determine proper weights among those factors in addition to optimizing the weight over the membership functions using a bat algorithm. The simulation results indicate the superior performance of the proposed method in terms of high throughput, low packet loss, balancing the overall energy consumption, and prolonging the network lifetime compared to state‐of‐the‐art protocols.