Real-time weapon detection is a critical component of intelligent surveillance systems, particularly for perimeter monitoring applications on embedded edge platforms. However, reliable alarm generation remains challenging because false positives, temporal instability, and viewpoint inconsistencies can propagate through conventional multi-camera fusion strategies. To address these limitations, this work proposes a lightweight Adaptive Multi-Camera Temporal Fusion (ACTF) framework that combines confidence-aware evidence separation, temporal persistence, and short-window cross-camera validation at the decision level, thereby confirming detections without requiring additional neural-network inference. The framework was evaluated using a TensorRT-optimized YOLO26s detector in controlled dual-camera scenarios involving clear visibility, partial occlusion, visually ambiguous distractors, and challenging illumination. While logical OR fusion achieved higher recall, it also propagated erroneous detections; in contrast, ACTF completely suppressed the distractor-induced false alarms while maintaining competitive performance and sub-second confirmation under favorable conditions. The original NVIDIA Jetson Nano implementation achieved an average throughput of 2.4 camera-pair cycles per second, corresponding to low-rate online embedded operation, whereas an additional NVIDIA Jetson Xavier NX benchmark achieved an average of 10.2 camera-pair cycles per second. This is equivalent to 10.2 processed frames per second for each camera stream and 20.4 camera images per second in aggregate. These results support reactive real-time embedded operation on the Xavier NX platform and demonstrate that ACTF improves alarm reliability with negligible decision-level computational overhead.
Carlos Julio Fierro-Silva, Carolina Del-Valle-Soto, S. M. Mostafa et al.· IEEE Access· 0 citations
(1) Background: Phishing remains a pervasive and economically damaging cyberthreat. The dominant detection paradigm has moved toward deep neural and transformer-based classifiers, a literature that reports high accuracy and that does not, in general, expose a per-decision justification, whereas interpretability and auditability are increasingly required in regulated environments; no comparison against transformer-scale detectors is made in this paper. This work asks how far a fully interpretable detector can close the accuracy gap to an opaque text classifier while preserving per-decision explanations, and what such a detector returns that accuracy alone does not measure. (2) Methods: HIEF, an interpretable evidence-fusion framework, is presented. Each email is represented by eighteen human-readable signals: fourteen structural and linguistic cues and four lexical aggregates derived from a published sparse log-odds lexicon. The signals are fused by three transparent layers, namely an L1-regularized logistic model, a shallow interaction-rule tree, and a calibrated Dempster–Shafer stage that reports belief, disbelief and ignorance masses together with an order-invariant global conflict coefficient derived in closed form. A logistic meta-learner fitted on out-of-fold component scores integrates the three layers. The evidential layer uses a type-aware calibration in which discrete signals are calibrated on their attainable values and continuous signals by isotonic regression. Evaluation uses 38,908 public emails, 38,512 of them after exact-duplicate removal, with near-duplicate control, group-aware partitioning, ten repeated splits, a source-held-out protocol, a two-class cross-source test set, a component ablation and a human audit of 100 messages annotated independently by two evaluators. (3) Results: Under group-aware partitioning, HIEF attains an F1 of 0.855 and the strongest term frequency–inverse document frequency (TF–IDF) baseline 0.954; a compact character n-gram neural reference model, evaluated over the same ten partitions, attains 0.973. The linear layer alone attains 0.872, so the two fusion layers do not improve accuracy over it, and the paired difference of 0.017 excludes zero. Type-aware calibration raises the evidential layer from 0.771 to 0.780 and more than halves its partition-to-partition standard deviation, but does not make it competitive; the weakness, therefore, lies in the fusion formulation rather than in the binning. What the evidential layer does supply is a decomposable account of decision uncertainty: the ignorance mass separates errors from correct decisions, 0.265 against 0.175. The human audit reaches an inter-annotator Cohen’s kappa of 0.950 over the five categories before adjudication, and shows that the permissive corpus label agrees with human phishing judgment at a Cohen’s kappa between 0.18 and 0.21, against 0.70 to 0.77 for the automatic strict rule; the audited block is annotated by two of the authors and its human positives are confined to the advance-fee family, so the audit is a bounded comparison of label assignments and not an independent annotation study. (4) Conclusions: HIEF is positioned as an uncertainty and explanation framework rather than as an accuracy-improving fusion method, since the measured accuracy cost of the fusion layers is not compensated by an accuracy gain. Quantifying how much of the performance reported on these widely used corpora is attributable to template leakage and to label permissiveness is a contribution independent of the detector itself. Cross-source operation has not been demonstrated: specificity falls to 0.041 on an unseen collection, so all evaluation reported here is proof-of-concept and no operational deployment claim is made. The Spanish-language evaluation rests on a small and entirely positive subset and is reported as preliminary.
Carolina Del-Valle-Soto, Carlos-Santiago Cruz-Diaz, M. Cardona et al.· Algorithms· 0 citations
Real-time weapon detection in video surveillance systems is a critical requirement for proactive security applications, particularly under the computational and latency constraints imposed by edge artificial intelligence deployments. While the YOLO family of object detectors has undergone continuous architectural evolution, the recently introduced YOLOv26 represents a significant redesign aimed at improving efficiency, stability, and deployment suitability across a wide range of hardware platforms. This work presents a comprehensive and homogeneous experimental evaluation of the full YOLOv26 model family, ranging from nano (YOLOv26n) to extra-large (YOLOv26x) variants, for real-time weapon detection in surveillance imagery. All models are trained and evaluated under identical conditions using a dataset that explicitly includes visually similar non-weapon objects as hard negatives, enabling a realistic assessment of false positives and false negatives in safety critical scenarios. The analysis encompasses training and validation dynamics, precision, recall evolution, mean Average Precision (mAP) at multiple IoU thresholds, class-wise confusion matrices, and inference latency. Results show that performance improves consistently from smaller to medium sized models, with YOLOv26m achieving the most balanced trade off between detection accuracy, robustness, and computational cost. Larger variants provide marginal accuracy gains at significantly higher complexity, revealing diminishing returns for edge oriented deployments. Overall, the findings demonstrate that the YOLOv26 architecture offers a scalable and mature detection framework, where model selection can be guided by explicit operational criteria rather than raw accuracy alone. This study establishes a strong baseline for future work on real world edge deployment, multi camera surveillance systems, and hardware aware optimization of next generation YOLO detectors.
Carlos Silva, Carolina Del-Valle-Soto, J. Varela-Aldás· 2026 6th International Confe...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
