<?xml version="1.0" encoding="utf-8"?>
    <rss version="2.0">
      <channel xmlns:content="http://purl.org/rss/1.0/modules/content/">
        <title>Frontiers in Computer Science | Computer Vision section | New and Recent Articles</title>
        <link>https://www.frontiersin.org/journals/computer-science/sections/computer-vision</link>
        <description>RSS Feed for Computer Vision section in the Frontiers in Computer Science journal | New and Recent Articles</description>
        <language>en-us</language>
        <generator>Frontiers Feed Generator,version:1</generator>
        <pubDate>2026-09-12T10:54:23.124+00:00</pubDate>
        <ttl>60</ttl>
        <item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1911640</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1911640</link>
        <title><![CDATA[Multi-stage generative enhancement and foundation model framework for automated dental implant identification: assessing the preprocessing paradox]]></title>
        <pubdate>2026-09-09T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Veena Benakatti</author><author>Ramesh P. Nayakar</author><author>Mallikarjun Anandhalli</author>
        <description><![CDATA[Artificial intelligence-based image analysis is transforming medical diagnosis; however, its clinical usefulness is limited by inherent noise and the quality of available medical datasets. Recent advances in generative AI promise to improve image quality through denoising, enhanced resolution, and more. This study applies the Segment Anything Model 2 (SAM2) and the Enhanced Super-Resolution Generative Adversarial Networks (ESRGAN) to dental radiographic datasets. A dataset of 6,413 radiographs across 35 dental implant systems was used to explore the pre-processing paradox, the relationship between human visual clarity and machine-learning accuracy. First, YOLOv11 detected and cropped implant areas, establishing the baseline for two classification models: ConvNeXt and ResNet50. These were compared with three experimental variants: SAM2 segmentation, ESRGAN super-resolution, and a fully integrated YOLO+SAM2 + ESRGAN pipeline. Performance was assessed using accuracy, F1-score, PSNR, and SSIM. In the baseline YOLO+CNN approach, the accuracy of ConvNeXt (83%) outperformed that of ResNet50 (79%). Adding SAM2 segmentation lowered performance (ConvNeXt: 81%, ResNet50: 76%), suggesting that removing the bone-implant interface eliminates important contextual signals. Both ConvNeXt and ResNet maintained their average accuracies (83 and 80%) when ESRGAN was applied to raw and cropped images. The largest difference appeared in the fully integrated pipeline, where ConvNeXt achieved higher accuracy (82%) compared to a significant decline in ResNet performance (64%). The study confirms the pre-processing paradox. While SAM2 and ESRGAN enhance visual clarity, the artifacts they introduce can reduce ResNet’s accuracy. The YOLOv11-CNN approach is optimal, and ConvNeXt is recommended for its robustness to image preprocessing. However, the proposed preprocessing pipeline did not significantly improve accuracy.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1908061</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1908061</link>
        <title><![CDATA[LPD-Net: lightweight adaptive weight-driven dual-stream network guided by physics-inspired priors for structure-preserving underwater image enhancement]]></title>
        <pubdate>2026-09-03T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Junming Liu</author><author>Jiajun Liang</author><author>Shaoliang Shi</author><author>Shiqiao Ruan</author><author>Junbo He</author><author>Qingnan Huang</author><author>Xiangsuo Fan</author>
        <description><![CDATA[IntroductionUnderwater visual sensing is severely hindered by wavelength-dependent absorption and scattering, which induce non-linear color shifts and structural haze. Existing enhancement methods often face a fundamental trade-off between chromatic restoration and geometric integrity, where color correction frequently occurs at the expense of blurring fine-grained details or amplifying noise.MethodsA lightweight physics-inspired dual-stream network (LPD-Net) is proposed for structure-preserving enhancement, establishing an explicit mapping from physical degradation phenomena to targeted neural operators. A dual-stream architecture is designed to decouple chromatic restoration and geometric reconstruction by utilizing a raw sensor stream as a fidelity anchor and a physics-inspired prior stream for guidance. A key innovation is the spatial-difference gated fusion mechanism, which dynamically optimizes the pixel-wise weights between raw sensor observations and empirical priors through data-driven perception. By employing end-to-end joint optimization, the network adaptively suppresses prior-induced artifacts in clear water while leveraging physics-inspired guidance in turbid scenarios.ResultsThe performance of the proposed model was evaluated using the UIEB and LSUI benchmarks. Experimental results indicate that LPD-Net achieves superior restoration quality, outperforming the fully aligned, competitive baseline by up to 0.67 dB in PSNR and securing the lowest perceptual distortion (0.1227 LPIPS). With a minimalist footprint of only 1.84 M parameters, the framework attains a core neural inference throughput of 27.70 FPS at 256 × 256 resolution on a mid-range GPU.ConclusionThe proposed LPD-Net provides a robust, efficient, and high-fidelity solution for structure-preserving underwater image enhancement. Its resolution-agnostic design and competitive throughput demonstrate significant potential for real-time deployment on autonomous underwater platforms and structural survey systems.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1853963</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1853963</link>
        <title><![CDATA[Tri-fusion deep learning model for early breast cancer screening using multimodal biomedical data]]></title>
        <pubdate>2026-09-03T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>J. Lilly Roseline Mary</author><author>R. Venkatesan</author><author>M. Mythily</author><author>D. Jasmine David</author>
        <description><![CDATA[To optimize the early detection of breast cancer, a strong integration of heterogeneous biomedical signals is essential to improve diagnostic reliability and reduce false negatives. This study proposes a Tri-Fusion deep learning model for timely breast cancer screening, which integrates White Blood Cell (WBC) morphological features, mammogram image embeddings, and gene mutation signatures within a unified multimodal framework. WBC features are extracted from peripheral blood smear images using a CNN-based morphologic encoder, mammographic representation is ensured by a pretrained deep convolutional foundation based on breast imaging information, and genomic details are modelled using a completely connected mutation-signature encoder deduced from a breast cancer–related gene panel. To capture cross-domain correlation, a dense categorization head is used for feature-level fusion. The proposed model uses three publicly available datasets: a WBC image dataset containing 12,500 samples, a CBIS-DDSM mammogram dataset containing 3,102 annotated instances, and a curated Genomic Dataset containing a mutant profile of 1200 patients from TCGA-BRCA. The experimental results show high accuracy compared with the unimodal and bimodal baselines, corresponding to 96.2% accuracy, 95.4% correctness, 94.8% recall, a F1-score of 95.1%, and an AUC-ROC of 98.6%. Furthermore, the model exhibits strong generalization under cross-validation and robustness to the missing-modality scenario.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1833609</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1833609</link>
        <title><![CDATA[Linearly constrained multi-representation domain generalization for unseen-bearing RUL prediction]]></title>
        <pubdate>2026-09-03T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Zhuoheng Dai</author><author>Lei Jiang</author><author>Liang Peng</author><author>Yingna Chen</author>
        <description><![CDATA[Accurate prediction of the remaining useful life (RUL) of rolling element bearings under target-bearing data scarcity remains a critical challenge in prognostics and health management (PHM). This paper proposes a multi-representation domain generalization framework for unseen-bearing RUL prediction. The framework uses a Bidirectional Multi-scale ConvLSTM architecture to capture temporal degradation dependencies and multi-granular spatial patterns from two aligned representations of the same vibration signal: the raw time-domain signal and its continuous wavelet transform (CWT) time-frequency representation. In addition, a lightweight linearly constrained temporal prior is integrated into the prediction layer to encourage a monotonically decreasing degradation trend, following the common damage-irreversibility assumption in bearing prognostics. The target bearing is excluded from training, validation, normalization-parameter estimation, and model selection, and is used only for final testing. Comprehensive experiments on two public benchmark datasets show that the proposed framework improves RUL prediction accuracy and provides competitive cross-bearing generalization.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1892207</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1892207</link>
        <title><![CDATA[Spatiotemporal vision transformers with Byzantine-robust federated prompt tuning for continuous urban perception]]></title>
        <pubdate>2026-08-31T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Rajesh Ankareddy</author><author>Radhakrishnan Delhibabu</author>
        <description><![CDATA[IntroductionWhile spatial Vision Transformers (ViTs) achieve high precision in urban scene parsing, their frame-by-frame application in autonomous driving suffers from severe temporal flickering and prohibitive retraining costs across decentralized vehicle fleets.MethodsTo overcome these dual bottlenecks, this paper introduces a unified Spatiotemporal Hierarchical Mask-Refinement (ST-HMR) framework integrated with a Byzantine-Robust Federated Learning (BR-FL) protocol. The ST-HMR module caches fine-grained prompt tokens via an asymmetrical Temporal Cross- Attention buffer to enforce inter-frame geometric continuity. Concurrently, the BR-FL pipeline employs Multi-Krum geometric distance filtration to aggregate 3 localized prompt gradients from decentralized fleets securely, updating only a 1.4% active parameter subset.ResultsEvaluated on the Cityscapes Video dataset, the ST-HMR framework improves the video segmentation mean Intersection over Union (mIoU) to 83.5%, elevates the Temporal Consistency (TC) score to 88.5, and reduces depth Absolute Relative Error (Abs Rel) to 0.085, all while maintaining real-time edge processing at 38 FPS. Under severe adversarial network conditions (up to 30% Byzantine/malicious sensor nodes), the BR-FL protocol achieves a 98.4% Byzantine detection rate and maintains a global mIoU of 81.9%, while reducing Over-The-Air (OTA) transmission payloads by over 99% (3.8 MB vs. 1.2 GB per round).DiscussionThese findings demonstrate that parameter-efficient prompt caching eliminates temporal boundary jitter without heavy 3D transformer overhead, while geometric gradient filtering provides robust defense against decentralized poisoning, establishing a scalable, secure, and temporally coherent perception paradigm for next-generation edge robotics.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1859574</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1859574</link>
        <title><![CDATA[A feature-enhanced and attention-focused intelligent detection method for water-floating garbage]]></title>
        <pubdate>2026-08-12T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Zhengwei Wang</author><author>Baijing Wu</author><author>Ke Gao</author>
        <description><![CDATA[To address the problems of low detection accuracy and large localization errors for small water-floating garbage targets under stacked occlusion, water surface fluctuations, and illumination variation, this study proposes a small-target water-floating garbage detection model named MFPF-YOLOv11s. Firstly, a multi-stage dilated residual module is designed to suppress interfering features caused by occluded targets through multi-scale network filtering, thereby enhancing the extraction of subtle garbage features. Secondly, the Meta-ACON activation function is introduced to improve the robustness of the model in capturing the nonlinear characteristics of water-floating garbage. Next, a Context Aggregation attention mechanism is embedded into the neck to guide the network to focus more effectively on garbage target regions. Finally, the bounding box regression loss is replaced with Wise-IoU. By exploiting its dynamic non-monotonic focusing mechanism, the proposed model further improves detection accuracy and localization precision. Experimental results on the Yellow River water-floating garbage dataset demonstrate that, compared with the baseline model, MFPF-YOLOv11s improves precision, recall, and mAP50 by 3.1, 3.2, and 2.3%, respectively, with only 0.06 M additional parameters and 0.7G additional FLOPs. These results indicate that the proposed method can meet the requirements of high-precision water-floating garbage detection and provide technical support for the intelligent monitoring and management of aquatic environments.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1838941</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1838941</link>
        <title><![CDATA[Semantic de-identification of burned-in PHI in DICOM medical images: a deep learning–NLP pipeline validated on clinical and phantom TMM datasets]]></title>
        <pubdate>2026-07-31T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Remya Sethulekshmi</author><author>Manu J. Pillai</author><author>Nihal Ahammed</author><author>Somula Ramasubbareddy</author><author>Yong Yun Cho</author>
        <description><![CDATA[IntroductionThe growing adoption of AI-based healthcare research has increased the need for properly anonymized medical imaging datasets. PHI within DICOM files - particularly burned-in pixel-level text - poses significant privacy and regulatory risks. Existing methods either focus solely on metadata or remove all detected text indiscriminately, sacrificing clinically relevant annotations.MethodsThis paper proposes a semantic de-identification pipeline integrating YOLOv11n-based text detection, domain-optimized EasyOCR, and a hybrid natural language processing (NLP) classification module combining regular expressions, keyword matching, and named entity recognition. A dual-path architecture processes metadata and pixel-level PHI in parallel, enabling complete DICOM sanitization while preserving non-PHI clinical annotations. The system was evaluated on 1,042 multi-modality DICOM images (CT, MRI, X-ray, ultrasound). As a secondary evaluation, the pipeline was also applied to two tissue-mimicking material (TMM) phantom datasets from TCIA - the RIDER Phantom MRI and Phantom FDA CT (RIDER = Reference Image Database to Evaluate Therapy Response; FDA = Food and Drug Administration) - which served as surrogates for controlled evaluation of metadata and burned-in identifier removal.ResultsThe system achieves an F1-score of 95.4%, 96.1% recall, a structural similarity index measure (SSIM) of 0.969, a peak signal-to-noise ratio (PSNR) of 28.9 dB, and processes each image in 2.8 s. It achieves SSIM of 0.986 and PSNR of 49.0 dB on RIDER Phantom MRI, and SSIM of 0.974 and PSNR of 31.5 dB on Phantom FDA CT.DiscussionThese results confirm that the pipeline preserves quantitative pixel fidelity when applied to institutional and device identifiers embedded in phantom acquisitions, supporting blinding for domain-generalization studies across institutions. The modular design supports institutional customisation, making it suitable for clinical research workflows and privacy-compliant phantom imaging pipelines.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1798168</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1798168</link>
        <title><![CDATA[Learning under extreme data scarcity: an enhanced hybrid CRNN with calibration and test-time augmentation for multi-class lung CT classification]]></title>
        <pubdate>2026-07-14T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>V. Maheswari</author><author>H. Parveen Sultana</author>
        <description><![CDATA[IntroductionThis study evaluates multi-class lung CT classification under extreme data scarcity.MethodsFrom a 400/72/295 train/validation/test split, only a class-balanced 5% coreset (20 images) is used for training. Six CNN baselines (ResNet-18, ResNet-50, DenseNet-121, MobileNetV3-Large, EfficientNet-B0, ConvNeXt-Tiny) are compared with a hybrid CRNN (EfficientNet-B0 encoder, bidirectional LSTMs, multi-head attention). Experiments are repeated over 3 seeds x 3 coreset samplings (nine runs per model), reported as mean +/- SD and validated with paired Wilcoxon signed-rank tests (Bonferroni correction); test-time augmentation (TTA) and post-hoc temperature scaling are assessed.ResultsThe proposed model attains accuracy 0.4523 +/- 0.0085, macro-F1 0.4637 +/- 0.0085, and macro-AUC 0.7302 +/- 0.0073, outperforming the best baseline ResNet-18 (0.3842 +/- 0.0089; p < 0.01, Cohen’s d = 7.89). TTA preserves accuracy (0.4557 +/- 0.0085) and gives the highest single-model macro-AUC (0.7357 +/- 0.0069); a top-3 ensemble reaches macro-AUC 0.7428 +/- 0.0072 but lowers accuracy (0.4105 +/- 0.0075). An ablation study attributes the largest gain to BiLSTM (+5.76%), with attention (+3.72%) and full combination (+10.00%). Farthest-point coreset selection yields higher mean accuracy and lower variance.DiscussionTemperature scaling is model-dependent; cross-seed analysis gives a stable T = 1.257 +/- 0.063, reducing ECE from 0.1557 +/- 0.0050 to 0.1190 +/- 0.0037, revising the original single-run finding.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1819991</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1819991</link>
        <title><![CDATA[A unified benchmarking framework for vector databases in scalable embedding-based image retrieval systems]]></title>
        <pubdate>2026-07-14T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Somula Ramasubbareddy</author><author>S. Remya</author><author>Sathishkumar Veerappampalayam Easwaramoorthy</author><author>Gaddam A. Sanjeeva Reddy</author><author>Yong Yun Cho</author><author>Manu J. Pillai</author>
        <description><![CDATA[Modern image retrieval and similarity search systems rely on high-dimensional embeddings that are produced by deep neural networks. To store and query these embeddings, vector databases are used. However, the lack of standardized, workload-aware benchmarks makes selecting the right system for a given workload challenging. This paper presents an adaptive benchmarking framework that systematically evaluates vector databases using realistic image-embedding workloads. The framework integrates CLIP ViT-B/32 (512-dimensional) and ResNet-50 embedding models with a consistent indexing and querying pipeline, thereby enabling fair and reproducible evaluation across the entire data lifecycle. We benchmarked four leading vector database systems–FAISS, Milvus, Qdrant, and pgvector–using CIFAR-10 and STL-10 image embeddings scaled from 100 to 10,000 vectors, with cross-dataset validation confirming that relative system rankings are consistent across different embedding distributions, with additional scalability tests up to 5 × 104 vectors. Our experiments reveal distinct performance trade-offs: FAISS inserts 10,000 vectors in 15 ms and answers queries in under one millisecond, but lacks built-in persistence. In contrast, pgvector, because of the transactional overhead typical of relational systems, needs 14.9 s for the same insertion task. Milvus balances 47–56 ms query latency with perfect Recall@5 across the lifecycle scales (100–10,000 vectors), with a small drop to 0.98–0.99 at the extended scalability scale of 50,000 vectors due to IVF cluster-boundary quantization error, while Qdrant, which offers the quickest update performance among persistent systems (approximately 2.7 ms across all tested scales), compromises on recall variability (0.88–1.00). At all tested scales, each of FAISS, Milvus, and pgvector scored perfect Recall@5. To support deployment decisions, a workload-aware multi-objective scoring model is suggested that considers latency, memory, and accuracy according to application requirements. The framework thus provides practical deployment insights for practitioners and also provides a reproducible foundation for vector database evaluation in AI-powered image retrieval.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1851511</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1851511</link>
        <title><![CDATA[DSF-Net: dual selective fusion network with spatial-frequency domain encoding for retinal vessel segmentation]]></title>
        <pubdate>2026-07-13T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Feng Liang</author><author>Xiaoqi Sheng</author><author>Yang Liu</author><author>Ruizhuo Li</author><author>Yanmin Shi</author>
        <description><![CDATA[IntroductionRetinal vessel segmentation is a fundamental task in quantitative fundus image analysis. However, existing methods still face challenges in segmenting thin and complex vessels because local details and global contextual information are often insufficiently integrated.MethodsTo address this issue, we propose a Dual Selective Fusion Network (DSF-Net) for retinal vessel segmentation. The proposed network consists of a Dual-Branch Encoder (DB-Encoder), a Pinwheel-based Local Attention (PLA) module, and a Dual Selective Fusion Transformer Block (DSFTB). The DB-Encoder jointly models spatial- and frequency-domain information to capture both fine vessel details and global contextual patterns. The PLA module enhances local perception and boundary sensitivity through asymmetric multidirectional convolutions and Sobel edge priors. The DSFTB integrates Multi-scale Feature Attention (MSFA) and token-selective Global Feature Attention (GFA) to enable adaptive feature fusion and long-range dependency modeling.ResultsExperiments conducted on the DRIVE, STARE, and CHASE_DB1 datasets demonstrate that DSF-Net achieves competitive overall performance compared with existing methods. In particular, the proposed method produces more accurate and structurally coherent segmentation results, especially for thin and complex vessels.DiscussionThese findings indicate that the combined modeling of local detail, frequency-aware representation, and global contextual dependency is effective for retinal vessel segmentation. DSF-Net provides a robust framework for improving vessel continuity and boundary delineation in fundus images. The source code is available at: https://github.com/liang050629/DSF-Net.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1837091</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1837091</link>
        <title><![CDATA[LightFormer-3D: a lightweight hierarchical CNN-transformer hybrid for 3D medical image segmentation]]></title>
        <pubdate>2026-07-09T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Qingzheng Hu</author><author>Ji Li</author><author>Wenqiu Zhu</author>
        <description><![CDATA[IntroductionBrain tumor segmentation from 3D MRI is critical for treatment planning, yet deploying accurate Transformer-based models in clinical settings remains challenging due to prohibitive computational costs. State-of-the-art methods often exceed 90M parameters and require seconds per inference, precluding real-time use on resource-constrained hardware.MethodsWe propose LightFormer-3D, a lightweight hierarchical CNN-Transformer hybrid that applies a co-design principle to jointly optimize patch embedding, self-attention, and multi-scale feature fusion, ensuring that efficiency gains compound across all stages. Specifically, a Depthwise Separable Convolution Patch Embedding (DSC-PE) reduces embedding parameters by 70%; a Multi-head Self-Attention with Spatial Sequence Reduction (MSA-SR) compresses self-attention complexity from O(N2) to O(N2/r3); and a Lightweight Scalable Feature Fusion (LSFF) module adaptively weights multi-scale features using only five parameters.ResultsEvaluated on the BraTS 2017 and BraTS 2021 benchmarks, LightFormer-3D achieves a mean Dice score of 87.0% on BraTS 2021 with 1.78M parameters and 9.3 GFLOPs, surpassing UNETR (80.2% Dice, 92.5M parameters) by 6.8 pp and the best competing lightweight method, U-Net_ASPP_EVO (86.0%), by 1.0 pp, while delivering 10× faster inference than UNETR (0.51s vs. 5.41s per case on an NVIDIA RTX 4060). Five-fold cross-validation (86.7% ± 0.2%) confirms stable generalization. Ablation studies validate each module's contribution, and zero-shot external validation on the independent MU-Glioma-Post dataset (203 post-treatment cases) confirms robust whole-tumor localization (WT Dice 85.0%) under severe domain shift.DiscussionLightFormer-3D establishes that co-designed lightweight modules can simultaneously improve accuracy and efficiency, achieving a new Pareto-optimal point for brain tumor segmentation.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1809919</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1809919</link>
        <title><![CDATA[DMFNet: exploring diverse mid-feature for visible-infrared person re-identification]]></title>
        <pubdate>2026-07-09T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ganqing Mo</author><author>Yanbing Chen</author><author>Hairong Ye</author>
        <description><![CDATA[IntroductionVisible-infrared person re-identification remains a challenging task due to inherent modality discrepancies between RGB and infrared images. Existing methods often struggle to effectively capture both modality-specific and modality-invariant features simultaneously, limiting their cross-modal matching performance.MethodsThis paper presents DMFNet (Diverse Mid-feature Network), a novel deep learning architecture that effectively harnesses intermediate shared features to bridge this cross-modal gap. DMFNet integrates two key modules: a Multi-layer Feature Cascade Module (MFCM) that aggregates discriminative features across different network stages, and a Dual Feature Generation Module (DFGM) that produces diverse intermediate representations through Instance-Batch Normalization variants.ResultsExtensive experiments on the SYSU-MM01 and RegDB datasets demonstrate that DMFNet achieves state-of-the-art performance, with significant improvements in Rank-1 accuracy (up to 8.2% on SYSU-MM01 and 6.5% on RegDB) and mean Average Precision (mAP) over existing methods.DiscussionOur approach not only enhances cross-modal matching capabilities but also provides interpretable feature visualizations, offering valuable insights into the network's decision-making process. These results pave the way for more robust person re-identification systems in real-world surveillance scenarios, particularly in low-light conditions where traditional visible-only systems often fail.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1707808</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1707808</link>
        <title><![CDATA[MoDiPO: text-to-motion alignment via AI-feedback-driven direct preference optimization]]></title>
        <pubdate>2026-07-03T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Massimiliano Pappa</author><author>Luca Collorone</author><author>Giovanni Ficarra</author><author>Indro Spinelli</author><author>Fabio Galasso</author>
        <description><![CDATA[Diffusion Models have revolutionized the field of human motion generation by offering exceptional generation quality and fine-grained controllability through natural language conditioning. Their inherent stochasticity, that is the ability to generate various outputs from the same input prompt, is key to their success. However, this diversity should not be unrestricted, as it may lead to unlikely generations. Instead, it should be confined within the boundaries of text-aligned and realistic generations. To address this issue, we propose MoDiPO (Motion Diffusion DPO), the first methodology to adapt Diffusion Direct Preference Optimization to align text-to-motion diffusion models. We streamline the laborious and expensive process of gathering human preferences needed in DPO by leveraging AI feedback instead. This enables us to experiment with novel DPO strategies, using both online and offline generated motion-preference pairs. To foster future research we contribute with a motion-preference dataset which we dub Pick-a-Move. We demonstrate, both qualitatively and quantitatively, that our proposed method yields significantly more realistic motions. In particular, MoDiPO achieves statistically significant improvements in Fréchet Inception Distance (FID) of up to 39% on MLD/HumanML3D and consistent gains of 9%–15% across both MLD and MDM on HumanML3D and KIT-ML. Finally, MoDiPO secures a threefold increase in preference from human evaluators compared to the original models' outputs.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1830781</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1830781</link>
        <title><![CDATA[Shot boundary detection for locating subliminal stimuli in films]]></title>
        <pubdate>2026-06-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Javier Sanz-Aznar</author><author>Juan José Caballero-Molina</author><author>Carlos Aguilar-Paredes</author>
        <description><![CDATA[IntroductionWith the objective of detecting film shots with a duration that would qualify them as subliminal, this paper proposes a shot boundary detection system to locate shots in a video that last for only one or two frames.MethodsThis design prioritizes the complete retrieval of extremely short shots, minimizing false negatives, even at the expense of additional false positives. This is achieved by a sequential filtering strategy, integrating complementary metrics in a cascade, ensuring robustness against motion, luminance variations and noise while reducing the processing time by progressively ruling out potential cases for comparison. The series of processes adopted were a color histogram analysis with a dynamic threshold, a mean squared error analysis (MSE), a structural similarity index measure (SSIM) and a histogram correlation analysis. The system design was tested on 10 films in which the shots that had to be detected were known beforehand, and subsequently on 15 other films that may potentially contain subliminal shots.ResultsThe results reveal the effective detection of 100% of the shots, although the erroneous detections produced require a visual review at the end of the process. Finally, the analysis was applied to 1,247 films to verify, in a real-world scenario, whether the designed system is effective in detecting one- or two-frame shots.DiscussionIt was thus determined that the design proposed can facilitate the identification of short shots for a large number of films, meeting the initial objective of the design and confirming the usefulness of this system for detecting subliminal stimuli in films.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1841416</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1841416</link>
        <title><![CDATA[Attention-guided hybrid CNN–transformer framework for cross-modality COVID-19 detection and lesion localization in CT and chest X-ray images]]></title>
        <pubdate>2026-06-10T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Manas Gautam</author><author>P. Jagalingam</author>
        <description><![CDATA[Accurate and interpretable analysis of thoracic medical images is essential for reliable COVID-19 diagnosis; however, existing approaches face significant challenges in jointly performing classification and lesion localization across multiple imaging modalities. Suboptimal diagnostic knowledge and limited spatial awareness are the results of conventional deep learning approaches, especially models based on convolutional neural networks (CNNs). These methods mainly concentrate on local feature extraction and often handle classification and segmentation as independent tasks. Although transformer-based models enhance global contextual learning, their practical use in real-world clinical settings is limited due to their high computing resource requirements and large-scale dataset requirements. On top of that, there are currently no explainability procedures built into hybrid CNN-Transformer methods, and most of them use shallow or one-directional fusion algorithms. In order to overcome these shortcomings, this research suggests a CNN-Transformer hybrid attention-guided paradigm for COVID-19 identification and lesion localization across modalities utilizing separate CT and CXR datasets. The suggested design incorporates a ResNet-18 backbone for local feature extraction, a light Vision Transformer for global contextual modelling, a bidirectional cross attention fusion module for effective interaction between CNN and transformer representations, a feature refinement mechanism based on convolutional block attention module (CBAM), and a unified multi-task learning framework. In the CT data set, the experimental results prove that the proposed model is superior to the other models, having an accuracy of 98.29%, a precision of 98.14%, a recall of 98.13%, and an F1-Score of 98.11%. The model also has a Dice score of 0.847 and an Intersection-over-Union score of 0.848, respectively. The proposed model gets an accuracy of 97.74%, a precision of 97.93%, a recall of 97.52%, and an F1-Score. By enhancing generalizability and clinical reliability across imaging modalities, our findings demonstrate that the suggested paradigm delivers accurate, robust, and interpretable COVID-19 diagnosis.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1753080</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1753080</link>
        <title><![CDATA[Hardware- and software-based fingerprint liveness detection schemes—A literature review]]></title>
        <pubdate>2026-06-02T00:00:00Z</pubdate>
        <category>Review</category>
        <author>Rubab Mehboob</author><author>Hassan Dawood</author><author>Ali Daud</author><author>Hussain Dawood</author>
        <description><![CDATA[Technological advancements have shaped the methods for ensuring personal safety and safeguarding digital assets. In this regard, biometric systems are innovative technological solutions that utilize various behavioral and physiological traits for personal verification and identification. This technology provides efficient and seamless authentication while improving users' convenience in a range of security applications, including border security, physical access control, healthcare, and financial transactions. Fingerprint-based verification systems are vulnerable to various attacks, including spoofing, presentation attacks, and latent fingerprint attacks. Henceforth, it is essential to incorporate a robust and reliable liveness detection mechanism into the authentication system to enhance the security of such systems. Since fingerprint liveness detection has been an active area of research, many techniques have been proposed for handling this problem by extracting different characteristics such as minutiae, ridges, texture, pores, perspiration, and morphological features. In this article, the state-of-the-art liveness detection schemes along with their taxonomy from 2003 to 2025 are chronologically analyzed. In this article, we have presented the techniques, their parameters, their pros and cons, and the evaluation matrices. Moreover, this article inculcates a comprehensive survey of the various hardware-, software-based approaches focusing on the deep learning liveness detection techniques. We have also investigated the standard benchmarks of the LivDet series that are publicly available and the evaluation protocols.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1789702</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1789702</link>
        <title><![CDATA[Comparative analysis of previous YOLO detectors and YOLOv26s for real-time weapon detection in video surveillance]]></title>
        <pubdate>2026-04-28T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Carlos Julio Fierro Silva</author><author>Carolina Del-Valle-Soto</author><author>Carlos Bran</author><author>José Varela-Aldás</author>
        <description><![CDATA[Real-time weapon detection in video surveillance is a critical capability for artificial intelligence assisted security systems, particularly in scenarios constrained by low latency, limited computational resources, and strict power efficiency requirements typical of edge artificial intelligence deployments. This work presents a comparative analysis of lightweight YOLO based object detectors, namely YOLOv8s, YOLOv9s, YOLOv10s, and YOLOv11s, vs. the recently introduced YOLOv26s model. In contrast to conventional benchmarking studies, this work extends the evaluation to real-world edge deployment conditions using an NVIDIA Jetson Nano device, explicitly measuring end-to-end latency, including preprocessing, inference, and post-processing stages. While earlier YOLO variants primarily relied on convolutional neural network architectures and intermediate explorations such as attention centered designs aimed to improve detection accuracy, YOLOv26 represents a paradigm shift by being designed from the ground up for low power edge devices, emphasizing architectural simplicity and deployment efficiency. To ensure a fair and reproducible evaluation, all models are trained on the same weapon detection dataset under a unified experimental protocol using small scale variants.The experimental results reveal that, despite exhibiting comparable inference times, different models show significantly different real-time performance due to variations in post-processing complexity. Specifically, models such as YOLOv8s, YOLOv9s, and YOLOv11s incur a substantial post-processing overhead, whereas YOLOv10s and YOLOv26s produce compact output representations that drastically reduce post-processing cost.This leads to a clear separation in deployment behavior, where end-to-end latency is reduced from approximately 300 ms to 125–130 ms, effectively doubling the achievable frame rate on embedded hardware. Rather than proposing a universal ranking, the study analyzes the trade offs introduced by architectural evolution and optimization strategies, providing technical criteria to support model selection under resource constrained deployment scenarios and demonstrating that post-processing efficiency, rather than inference speed alone, is the dominant factor in real-time edge performance.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1824259</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1824259</link>
        <title><![CDATA[Correction: An improved contrastive learning loss function for automated clock-drawing test grading with implications for cognitive impairment screening]]></title>
        <pubdate>2026-03-31T00:00:00Z</pubdate>
        <category>Correction</category>
        <author>Ning Liu</author><author>Qian Sun</author><author>Xiaoyin Xu</author><author>Haifeng Mou</author><author>Xinhai Liao</author><author>Bokai Rong</author><author>Lingxing Wang</author>
        <description></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1753764</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1753764</link>
        <title><![CDATA[Single-item training for multi-dish recognition: a class-agnostic framework for Indian food platters]]></title>
        <pubdate>2026-03-10T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Keerthi Garisa</author><author>Ravi Kant Kumar</author><author>Priyanka Singh</author>
        <description><![CDATA[Accurate dietary assessment is increasingly dependent on automated food recognition systems capable of operating effectively in real-world environments. While most vision-based models perform well on single-item datasets, their performance degrades significantly in complex multi-dish settings. This scenario is particularly evident in Indian thalis, which contain overlapping food items with diverse textures and high visual variability. These challenges make large-scale multi-dish annotation expensive and limit practical deployment of such systems. To address this gap, we propose a novel two-stage framework that enables recognition of multi-dish food images using only single-item training data. The proposed pipeline incorporates class-agnostic segmentation using the Segment Anything Model (SAM), followed by classification with an SE-DenseNet121 network optimized via Optuna-based hyperparameter tuning.The model is trained exclusively on single-item annotated images and generalizes to multi-item thali images at inference time through a segmentation-classification mapping strategy. This zero-shot segmentation approach eliminates the need for multi-dish ground-truth annotations. As a result, the annotation complexity is reduced from O(N × M) to O(N). The proposed system achieves accuracy of 97.48% on single-item food image classification and demonstrates strong applicability to multi-dish Indian thali images through region-wise inference on segmented food items. Furthermore, the framework is computationally efficient, achieving 2 × faster inference with a latency of 1.58 ms while using only 70% of the parameters required by transformer-based baselines. It operates with low computational cost (2.90 GFLOPs), significantly fewer parameters (8.06M compared to 26.69–86.77M), and delivers higher throughput (633.32 samples/s). These results demonstrate that the proposed method provides a scalable and practical solution for real-time dietary assessment applications.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1690044</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1690044</link>
        <title><![CDATA[An improved contrastive learning loss function for automated clock-drawing test grading with implications for cognitive impairment screening]]></title>
        <pubdate>2026-02-20T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ning Liu</author><author>Qian Sun</author><author>Xiaoyin Xu</author><author>Haifeng Mou</author><author>Xinhai Liao</author><author>Bokai Rong</author><author>Lingxing Wang</author>
        <description><![CDATA[Contrastive learning has been attracting much interest in recent years for its ability to train without labeled data. An important factor in its success is the loss function, which guides the search for prominent features that separate the positive and negative classes. The triplet loss function is widely used in contrastive learning, in which the objective is to attract a pair of positive instances while pushing away a negative instance from the anchor instance, where one of the positive instances is often an augmented version of the anchor. To improve the performance of contrastive learning in automated Clock-Drawing Test (CDT) grading, this paper proposes a more comprehensive triplet loss function that aims to not only keep the distance between the anchor and a positive instance small and the distance between the anchor and a negative instance large, but also keep the distance between the positive and negative instances large. Experimental results show that the improved loss function significantly improves the model’s accuracy, precision, recall, and F1-score by 3–5% on both CIFAR-10 and CDT datasets, providing a new method for improving the accuracy of automatic CDT scoring and early detection of cognitive impairments.]]></description>
      </item>
      </channel>
    </rss>