Explore the vast range of titles available.

The high pace emergence in advanced software systems, low-cost hardware and decentralized cloud computing technologies have broadened the horizon for vision-based surveillance, monitoring and control. However, complex and inferior feature learning over visual artefacts or video streams, especially under extreme conditions confine majority of the at-hand vision-based crowd analysis and classification systems. Retrieving event-sensitive or crowd-type sensitive spatio-temporal features for the different crowd types under extreme conditions is a highly complex task. Consequently, it results in lower accuracy and hence low reliability that confines existing methods for real-time crowd analysis. Despite numerous efforts in vision-based approaches, the lack of acoustic cues often creates ambiguity in crowd classification. On the other hand, the strategic amalgamation of audio-visual features can enable accurate and reliable crowd analysis and classification. Considering it as motivation, in this research a novel audio-visual multi-modality driven hybrid feature learning model is developed for crowd analysis and classification. In this work, a hybrid feature extraction model was applied to extract deep spatio-temporal features by using Gray-Level Co-occurrence Metrics (GLCM) and AlexNet transferrable learning model. Once extracting the different GLCM features and AlexNet deep features, horizontal concatenation was done to fuse the different feature sets. Similarly, for acoustic feature extraction, the audio samples (from the input video) were processed for static (fixed size) sampling, pre-emphasis, block framing and Hann windowing, followed by acoustic feature extraction like GTCC, GTCC-Delta, GTCC-Delta-Delta, MFCC, Spectral Entropy, Spectral Flux, Spectral Slope and Harmonics to Noise Ratio (HNR). Finally, the extracted audio-visual features were fused to yield a composite multi-modal feature set, which is processed for classification using the random forest ensemble classifier. The multi-class classification yields a crowd-classification accurac12529y of (98.26%), precision (98.89%), sensitivity (94.82%), specificity (95.57%), and F-Measure of 98.84%. The robustness of the proposed multi-modality-based crowd analysis model confirms its suitability towards real-world crowd detection and classification tasks.

Unmanned aerial vehicles (UAVs), commonly known as drones, have emerged as valuable tools in disaster response due to their ability to access remote and hard-to-reach areas. Equipped with camera sensors, UAVs facilitate real-time monitoring of disaster-stricken regions, including collapsed buildings, floods, and fires, enabling faster mitigation efforts. However, integrating deep learning (DL) models into UAVs for disaster detection introduces significant computational overhead, limiting their use in low-latency scenarios where rapid decision-making is critical. To address this challenge, this paper proposes a novel model for precise emergency spot detection that combines feature extraction from DenseNet, hyperparameter tuning using Penguin Search Optimization (PESO), and classification through an Augmented Feature Regional Convolutional Neural Network (AF-RCNN). The model introduces several innovations, including the Small Spots Feature Enrichment Extractor (SSFEE) for enhanced detection of small-scale disaster spots, the SCM Loss function for balancing feature uniqueness and commonality across varying spot sizes, and a one-to-one computation approach for optimized data sampling. Furthermore, Duck Swarm Optimization (DSOA) is employed to fine-tune the AF-RCNN parameters. The proposed model is evaluated using the AIDER dataset, and experimental results demonstrate superior performance compared to existing models, achieving faster processing speeds—up to 20 times quicker—while maintaining or surpassing accuracy. These findings highlight the potential of the proposed model for real-time, efficient emergency detection in disaster response scenarios.

The exponential rise in technology and allied applications has always revitalized academia industries to achieve more efficient and robust solution to meet contemporary demands. Surveillance systems have always been the dominant area which has grabbed the attention of the scientific community to enable real-time events or target’s characterization to make timely decision process. Crowd behavior analysis and classification is one of the most sought, though complex system to meet at hand surveillance purposes. However, unlike pedestrian movement detection methods, crowd analysis and behavioral characterization require robust feature learning and classification. With this motive, in this paper, a highly robust model is developed by applying hybrid deep features containing statistical features of the gray-level co-occurrence matrix (GLCM) and transferable deep learning AlexNet high-dimensional features. In addition, to perform multi-class classification multi-feed forward neural network model (MFNN) is used. Here, the inclusion of hybrid features of GLCM and AlexNet provides deep spatio-temporal feature information which helps in making optimal classification decision. On the other hand, the use of MFNN algorithm enables optimal multi-class classification. Thus, the combined model with hybrid deep features and MFNN achieves crowd behavior classification with 91.35% accuracy, 89.92% precision, 88.34% recall and F-measure of 89.12%.

As education institutions face new security challenges, the integration of Artificial intelligence (AI) and computer vision with surveillance systems for real time monitoring and threat detection is becoming mainstream. The inefficiency of a traditional CCTV system, which rely on human monitoring, makes them susceptible to costly mistakes. This literature survey examines deep learning methodologies focused on Convolutional Neural Networks, YOLO based object detection, Haar Cascade classification, and Local Binary Pattern Histogram for campus surveillance and recognition systems used in the 46 works collected between 2020 and 2025. The survey tracks the advancements made towards systems that autonomously monitor and recognize faces, track vehicles, analyse crowds, and even detect behaviours, as AI systems attain the ability to automate processes. Although the systems in question boost recognition accuracy exceeding 95%, real time system flexibility, varying lighting conditions, occlusions, privacy, and system scalability remind touchy problems. The survey suggests that the AI powered systems of the future should work towards smart frameworks that integrate disparate surveillance systems and automated alert systems.

This paper presents a comprehensive survey of automated machine learning (AutoML) techniques, with a focus on meta-learning, hyperparameter optimization, and neural architecture search (NAS). Rather than proposing or evaluating a full AutoML platform, this work synthesizes insights from 49 influential research papers, organizing them into methodological categories and highlighting their contributions to the evolution of ML automation. The survey also analyzes trends across model types, optimization strategies, and publication patterns. Based on this review, the paper identifies research gaps and outlines key directions for future development of unified, scalable, and interpretable AutoML systems. This survey is intended to provide a structured foundation for researchers working toward improved ML automation pipelines.

The dramatic ascent in advancements and associated applications have generally revived the scholarly community enterprises to accomplish more effective and strong answer to satisfy contemporary needs. Reconnaissance frameworks have drawn in mainstream researchers to empower occasions to settle on convenient choice interaction. This paper proposes a robust model to perform multi-class classification using Evolutionary Computing algorithm named Binary Bat Algorithm based Multi-Feed Forward Neural Network model (BBA-MFNN). Here, the utilisation of GLCM and AlexNet features provides deep spatio?temporal feature information helps to make optimal classification decision and the proposed BBA-MFNN algorithm enables optimal multi-class classification while avoiding local minima and convergence problem which can often be present in video data analysis due to non-linear feature distribution. Thus, the proposed model accomplishes Crowd Behaviour analysis with an accuracy of 96.15%, precision of 94.66%, 96.52% recall and F-Measure of 95.56%, which is higher than the classical MFNN classifier

Rapidly emerging SARS-CoV-2 variants jeopardize antibody-based countermeasures. Although cell culture experiments have demonstrated a loss of potency of several anti-spike neutralizing antibodies against variant strains of SARS-CoV-2 1 – 3 , the in vivo importance of these results remains uncertain. Here we report the in vitro and in vivo activity of a panel of monoclonal antibodies (mAbs), which correspond to many in advanced clinical development by Vir Biotechnology, AbbVie, AstraZeneca, Regeneron and Lilly, against SARS-CoV-2 variant viruses. Although some individual mAbs showed reduced or abrogated neutralizing activity in cell culture against B.1.351, B.1.1.28, B.1.617.1 and B.1.526 viruses with mutations at residue E484 of the spike protein, low prophylactic doses of mAb combinations protected against infection by many variants in K18-hACE2 transgenic mice, 129S2 immunocompetent mice and hamsters, without the emergence of resistance. Exceptions were LY-CoV555 monotherapy and LY-CoV555 and LY-CoV016 combination therapy, both of which lost all protective activity, and the combination of AbbVie 2B04 and 47D11, which showed a partial loss of activity. When administered after infection, higher doses of several mAb cocktails protected in vivo against viruses with a B.1.351 spike gene. Therefore, many—but not all—of the antibody products with Emergency Use Authorization should retain substantial efficacy against the prevailing variant strains of SARS-CoV-2. Experiments in mouse and hamster models show that monoclonal antibody combinations, using antibodies that correspond to products in clinical development, largely retain their efficacy in protecting against currently prevailing variant strains of SARS-CoV-2.

Gastric cancer is among the most prevalent and deadliest of cancers globally. To derive mechanistic insight into the pathways governing this disease, we generated a Claudin18-IRES-CreERT2 allele to selectively drive conditional dysregulation of the Wnt, Receptor Tyrosine Kinase and Trp53 pathways within the gastric epithelium. This resulted in highly reproducible metastatic, chromosomal-instable-type gastric cancer. In parallel, we developed orthotopic cancer organoid transplantation models to evaluate tumour-resident Lgr5 + populations as functional cancer stem cells via in vivo ablation. We show that Cldn18 tumours accurately recapitulate advanced human gastric cancer in terms of disease morphology, aberrant gene expression, molecular markers and sites of distant metastases. Importantly, we establish that tumour-resident Lgr5 + stem-like cells are critical to the initiation and maintenance of tumour burden and are obligatory for the establishment of metastases. These models will be invaluable for deriving clinically relevant mechanistic insights into cancer progression and as preclinical models for evaluating therapeutic targets. Fatehullah et al. develop transgenic and orthotopic mouse models to recapitulate advanced human gastric cancer and uncover a mechanistic role for Lgr5 + stem-like cells in promoting disease initiation and progression.

Antibodies are a principal determinant of immunity for most RNA viruses and have promise to reduce infection or disease during major epidemics. The novel coronavirus SARS-CoV-2 has caused a global pandemic with millions of infections and hundreds of thousands of deaths to date 1 , 2 . In response, we used a rapid antibody discovery platform to isolate hundreds of human monoclonal antibodies (mAbs) against the SARS-CoV-2 spike (S) protein. We stratify these mAbs into five major classes on the basis of their reactivity to subdomains of S protein as well as their cross-reactivity to SARS-CoV. Many of these mAbs inhibit infection of authentic SARS-CoV-2 virus, with most neutralizing mAbs recognizing the receptor-binding domain (RBD) of S. This work defines sites of vulnerability on SARS-CoV-2 S and demonstrates the speed and robustness of advanced antibody discovery platforms. A platform for rapid antibody discovery enabled the isolation of hundreds of human monoclonal antibodies against SARS-CoV-2 and the prioritization of potent antibody candidates for clinical trials in patients with COVID-19.

The integrated stress response (ISR) enables cells to cope with a variety of insults, but its specific contribution to downstream cellular outputs remains unclear. Using a synthetic tool, we selectively activate the ISR without co-activation of parallel pathways and define the resulting cellular state with multi-omics profiling. We identify time- and dose-dependent gene expression modules, with ATF4 driving only a small but sensitive subgroup that includes amino acid metabolic enzymes. This ATF4 response affects cellular bioenergetics, rerouting carbon utilization towards amino acid production and away from the tricarboxylic acid cycle and fatty acid synthesis. We also find an ATF4-independent reorganization of the lipidome that promotes DGAT-dependent triglyceride synthesis and accumulation of lipid droplets. While DGAT1 is the main driver of lipid droplet biogenesis, DGAT2 plays an essential role in buffering stress and maintaining cell survival. Together, we demonstrate the sufficiency of the ISR in promoting a previously unappreciated metabolic state. ISR-specific contributions to stress-induced cellular outputs are not well understood. Here, authors use a minimal activation system and multi-omics to define an ISR-sufficient metabolic state that includes protective accumulation of lipid droplets.