Digital Library

Vol. 21, No. 5, Oct. 2025  Cover

1. Application of Data Mining Technology in the Selection of Teaching Evaluation Indicators and the Construction of Teaching Information Evaluation Model in Colleges and Universities
   Jiangxia Han
   Vol. 21, No. 5, pp. 457-470, Oct. 2025  
   https://doi.org/10.3745/JIPS.04.0358
   Details Full Text PubReader
   Keywords: Apriori, BPNN, Data Mining, SSA, Teaching Evaluation
   Show / Hide Abstract

   Due to the low efficiency and poor accuracy of current college teaching intelligence evaluation methods, an improved method is proposed. Firstly, an improved apriori (IApriori) algorithm is utilized to filter evaluation indexes and establish a teaching quality evaluation indicator system. Secondly, considering the high complexity and low accuracy of the backpropagation neural network (BPNN), principal component analysis (PCA) is taken to reduce the input data’s dimension. An improved sparrow search algorithm (ISSA) is simultaneously utilized to optimize the parameters of BPNN. Finally, a PCA-ISSA-BPNN teaching intelligence evaluation model is constructed. The experiments validated that when the number of transactions was 1,000, the IApriori only took 0.32 seconds to run. While the number of projects was 11, IApriori ran in 15.28 seconds. The evaluation accuracy of the PCA-ISSA-BPNN model reached 99.05%, the F1 value was 96.43%, the recall was 97.26%, and the AUC was 0.981. The above data show that IApriori has a higher efficiency in data mining and can more effectively screen evaluation indicators. This research method can effectively and accurately evaluate teaching quality, and has a positive impact on promoting student development, advancing teaching reform, and improving teaching quality.


2. Design of In-Vehicle Life Detection System Based on FMCW Millimeter-Wave Radar
   Weiquan Li, Li Li, Songlin Wei, Ying Wang
   Vol. 21, No. 5, pp. 471-483, Oct. 2025  
   https://doi.org/10.3745/JIPS.04.0359
   Details Full Text PubReader
   Keywords: ECU (Engine Control Unit), FMCW (Frequency Modulated Continuous Wave), Millimeter-Wave Radar
   Show / Hide Abstract

   Remote non-contact monitoring of human vital signs has recently received lots of attention due to safety requirements. Among a variety of sensors available, millimeter-wave radars show great advancement and advantages. With the miniaturization of radar systems, frequency modulated continuous wave (FMCW) millimeter-wave radar has been employed to address the issue of in-vehicle life detection. In the designed in-vehicle life detection system, Yosemite 4T8R 77/79 GHz radar chip communicates with the engine control unit (ECU). The system can detect the availability of child in car, alarm and perform self-check. It is shown that the designed system has high accuracy and anti-interference ability. It has convenient interface with smart cockpit system and is suitable for a wide range of vehicles.


3. Improvement of Object Classification Performance Using a Fusion Optimizer
   Si-Ung Kim, Nammee Moon
   Vol. 21, No. 5, pp. 484-493, Oct. 2025  
   https://doi.org/10.3745/JIPS.02.0226
   Details Full Text PubReader
   Keywords: Convolution Neural Network, Image Classification, Optimizer Fusion, Vision Transformer
   Show / Hide Abstract

   Training deep learning models involves the use of various optimization algorithms, each with its own advantages and disadvantages. Stochastic gradient descent (SGD) provides consistent performance and stable optimization but has the drawback of a slow convergence rate. On the other hand, Adam offers the advantage of fast convergence but can lead to overfitting. This study proposes a hybrid method that combines the stable convergence of SGD with the fast convergence of Adam, enabling the model to be optimized quickly and stably. This approach was applied to the EfficientNetV2 and Vision Transformer (ViT) architectures in image classification tasks. EfficientNetV2 used Adam up to the 6th block and then switched to SGD, achieving the best performance on the proposed dataset with an accuracy of 97.84%, a loss of 0.0990, and an F1-score of 98.04%. Similarly, ViT used Adam for the first 10 encoders and then switched to SGD for the remaining 10 encoders, showing optimal results on the same dataset with an accuracy of 98.54%, a loss of 0.1345, and an F1-score of 98.53%. This fusion optimizer approach effectively enhances training by using Adam for initial feature extraction and SGD for later stages.


4. Application of Virtual Simulation in Traditional Chinese Medicine Processing Education
   Qi Zhang, Zixi Song, Huaying Zhou
   Vol. 21, No. 5, pp. 494-507, Oct. 2025  
   https://doi.org/10.3745/JIPS.04.0360
   Details Full Text PubReader
   Keywords: Realism and Immersion, Traditional Chinese Medicine Processing Experiment, Virtual Simulation
   Show / Hide Abstract

   Traditional Chinese medicine (TCM) processing is a highly practical discipline, and experimental teaching is an important component of the curriculum. However, the experimental teaching methods often face the limitations of seasonal time, experimental materials, and safety factors, resulting in fewer opportunities for students to practice, insufficient learning interest, and poor teaching effectiveness. Therefore, integrating virtual simulation into experimental teaching is an important means of solving the experimental teaching problem. In this paper, a virtual laboratory platform for TCM processing is built based on the principles of TCM processing, providing learners with a real experience in a TCM processing environment. Aiming at the problem of realism and immersion in virtual scenes, we present a series of key technologies, including using Maya for 3D model construction, physically based rendering technology for texture mapping, and Unity3D for complete interactive functions. The result shows that our developed platform has good performance, high immersion, and accurate interaction. Learners can roam through the scene, visualize the herb model, and complete experiments according to the prompted steps. Moreover, the virtual laboratory can deepen students’ understanding of the basic theories and enhance their learning outcomes, including knowledge, skills and enjoyment.


5. Single-Object Fall Detection Using Pose Estimation 3D Coordinates via Generating 3D Object Coordinates from 2D Object Coordinates
   Jinmo Yang, Kidu Kim, R. Young Chul Kim
   Vol. 21, No. 5, pp. 508-516, Oct. 2025  
   https://doi.org/10.3745/JIPS.02.0227
   Details Full Text PubReader
   Keywords: Dimensional Lifting, Pose Estimation, Object Tracking Window
   Show / Hide Abstract

   Object fall detection is one of the significant applications in pose estimation. Traditional approaches heavily rely on fully mature neural networks, which may be complex to implement and resource-intensive. In this paper, we suggest a simple approach to detect Human fall on a single object using a mathematical comparison mechanism from 3D pose estimation landmarks lifted from 2D landmarks, suitable for low-specification systems. Our research focuses on dimensional lifting in 2D to 3D pose estimation based on object tracking—to adapt mapping 2D toons to 3D toons. For future research, we aim to develop a lightweight neural network for enhanced performance from ensemble effects and complex action detection. We also plan on enhancing the algorithm for multi-person detection. At this moment, our research extends into 3D toon generation. Our method achieves an F1-score of 94.7% (accuracy of 96.7%) with 28.2 FPS in detecting falls from webcam footage in a controlled environment.


6. DFPINet: Dual Feature and Progressive Interaction Network for Point Cloud Registration
   Xinhua Lu, Hui Wan, Lingxiao Zhang, Hao Zhang, Zheng He
   Vol. 21, No. 5, pp. 517-530, Oct. 2025  
   https://doi.org/10.3745/JIPS.02.0228
   Details Full Text PubReader
   Keywords: Dual Feature, Partial-to-Partial, Point Cloud Registration, Progressive Feature Interaction
   Show / Hide Abstract

   In point cloud registration, the combination of global and local features provides a comprehensive representation of point cloud data. Previous registration methods predominantly relied on a single type of feature. For instance, correspondence matching-based methods require stringent criteria for the uniqueness of input point clouds, and the quality of feature extraction significantly influences the accuracy of registration results. Global feature-based methods capture the overall geometric information of the inputs; however, the absence of local information often leads to the neglect of the adverse effects caused by non-overlapping points. In this paper, we present DFPINet, an iterative network based on dual feature extraction and progressive feature interaction for partial-to-partial point cloud registration of small-scale object point clouds. We use a dual branch structure to individually extract both global and local features, fully utilizing the geometric characteristics of the point cloud. Additionally, the progressive feature interaction operation enhances feature connections and mitigates the effect of non-overlapping points. Experimental results demonstrate that our method surpasses existing registration approaches in both accuracy and robustness.


7. Instantaneous Frequency Estimation of Nonstationary Jammer Based on TVAR for DSSS
   Na Liang, Jining Feng
   Vol. 21, No. 5, pp. 531-541, Oct. 2025  
   https://doi.org/10.3745/JIPS.03.0209
   Details Full Text PubReader
   Keywords: Basis Function, DSSS, Instantaneous Frequency Estimation, Parameter Estimation, TVAR
   Show / Hide Abstract

   Direct-sequence spread-spectrum receivers are sensitive to nonstationary jammers, and instantaneous frequency (IF) estimation is an important component of antijamming. Time-varying autoregressive (TVAR) parametric modeling time–frequency analysis avoids the limitation of resolution by observation time and has the advantage of high time–frequency resolution. However, the performance of the TVAR model for IF estimation is affected by such factors as basis functions. To address this problem, the TVAR model of nonstationary signals was studied, and the performances of TVAR models with different basis functions for IF estimation of linear frequency modulation (LFM) and nonlinear frequency modulation (NLFM) jammers were investigated. Conclusions were drawn, and the optimal basis for the TVAR model used for IF estimation of LFM and NLFM jammers was obtained. The results provide a basis for applying of the TVAR model to nonstationary jammers.


8. Classification Model of Diabetic Retinopathy Based on a Lightweight Feature-Enhanced Residual Swin Transformer
   Na Li, Kai Ren
   Vol. 21, No. 5, pp. 542-554, Oct. 2025  
   https://doi.org/10.3745/JIPS.02.0229
   Details Full Text PubReader
   Keywords: Classification of Diabetic Retinopathy, Depthwise Separable Convolution, Residual Connection, Swin Transformer
   Show / Hide Abstract

   The automated detection of diabetic retinopathy (DR) relies heavily on retinal image analysis. While artificial intelligence models have shown promise in DR management, they often face challenges such as high computational complexity, reduced accuracy due to class imbalance and small inter-class gaps, and increased processing times. Addressing these limitations, this study introduces the lightweight feature-enhanced residual Swin (LFRS) Transformer, a model that maintains high accuracy despite significantly lowered computational demands. Our approach begins by converting color fundus images to grayscale, followed by local feature extraction using a depthwise separable convolution module. These features are subsequently subjected to processing by a lightweight Swin Transformer enhanced with residual connections, improving both global feature extraction and computational efficiency. Evaluated on the DR classification dataset released by APTOS 2019, the LFRS Transformer achieves an accuracy of 0.928, a recall of 0.965, and a weighted kappa score of 0.957. Compared to the baseline Swin Transformer, our model reduces computational load by 22.2 GFLOPs and decreases model parameters by 7.2M, demonstrating a substantial improvement in efficiency. These results underscore the LFRS Transformer as a highly efficient and reliable DR screening tool, positioning it as well-suited for large-scale clinical screening programs.


9. V-TRACE: Persistent Tracking Misbehaving Vehicles under Pseudonym Changes in V2X Networks
   Juntaek Lee, Jiwoo Lee, Chanmin Kim, Jiwon Seo
   Vol. 21, No. 5, pp. 555-563, Oct. 2025  
   https://doi.org/10.3745/JIPS.03.0210
   Details Full Text PubReader
   Keywords: Cooperative Intelligent Transportation Systems (C-ITS), Misbehaving Vehicle Tracking, V2X
   Show / Hide Abstract

   As connected and cooperative intelligent transportation systems (C-ITS) advance with vehicle-to-everything (V2X) communication, the risk of cyber threats such as false data injection and message flooding continues to grow. We present V-TRACE, a lightweight and standard-compliant framework for vehicle tracking and identification that operates effectively even under frequent pseudonym changes. V-TRACE initiates real-time tracking upon detecting anomalous behavior and correlates spatiotemporal patterns across V2X messages to sustain traceability. When identifier rotation occurs, the system seamlessly transitions to an identification phase that maintains persistent monitoring and suppresses further impact. Unlike approaches that require protocol modifications, V-TRACE operates fully within existing V2X standards, ensuring compatibility with current C-ITS deployments. We implement and evaluate V-TRACE using a high-fidelity CANoe.Car2x simulation. Experimental results demonstrate that V-TRACE reliably tracks and re-identifies misbehaving vehicles with minimal overhead, confirming its scalability and practicality in real-world traffic scenarios.