Publications

@article{sapkota_object_2026,

title = {Object detection with multimodal large vision-language models: An in-depth review},

author = {Ranjan Sapkota and Manoj Karkee},

url = {https://www.sciencedirect.com/science/article/pii/S1566253525006475},

doi = {10.1016/j.inffus.2025.103575},

issn = {1566-2535},

year  = {2026},

date = {2026-02-01},

urldate = {2026-02-01},

journal = {Information Fusion},

volume = {126},

pages = {103575},

abstract = {The fusion of language and vision in large vision-language models (LVLMs) has revolutionized deep learning-based object detection by enhancing adaptability, contextual reasoning, and generalization beyond traditional architectures. This in-depth review presents a structured exploration of the state-of-the-art in LVLMs, systematically organized through a three-step research review process. First, we discuss the functioning of vision language models (VLMs) for object detection, describing how these models harness natural language processing (NLP) and computer vision (CV) techniques to revolutionize object detection and localization. We then explain the architectural innovations, training paradigms, and output flexibility of recent LVLMs for object detection, highlighting how they achieve advanced contextual understanding for object detection. The review thoroughly examines the approaches used in integration of visual and textual information, demonstrating the progress made in object detection using VLMs that facilitate more sophisticated object detection and localization strategies. Furthermore, this review presents comprehensive visualizations demonstrating LVLMs’ effectiveness in diverse scenarios including localization and segmentation, and then compares their real-time performance, adaptability, and complexity to traditional deep learning systems. Based on the review analysis, its is expected that LVLMs will soon meet or surpass the performance of conventional methods in object detection. However, because of the unique and complimentary characteristics of traditional deep learning approaches and LVLMS, it is anticipated that hybrid approaches integrating both types of object detection models will be utilized in the future to maximize the speed, reliability and robotiness of the systems. Moreover, the review also identifies a few major limitations of the current LVLM modes, proposes solutions to address those challenges, and presents a clear roadmap for the future advancement in this field. We conclude, based on this study, that the recent advancement in LVLMs have made and will continue to make a transformative impact on object detection and automated applications in the future.},

keywords = {Information fusion, Language and vision fusion, Large language models, Object detection, Vision-language models},

pubstate = {published},

tppubtype = {article}

}

@article{raza_trism_2026,

title = {TRiSM for Agentic AI: A review of Trust, Risk, and Security Management in LLM-based Agentic Multi-Agent Systems},

author = {Shaina Raza and Ranjan Sapkota and Manoj Karkee and Christos Emmanouilidis},

url = {https://www.sciencedirect.com/science/article/pii/S2666651026000069},

doi = {https://doi.org/10.1016/j.aiopen.2026.02.006},

issn = {2666-6510},

year  = {2026},

date = {2026-01-01},

urldate = {2026-01-01},

journal = {AI Open},

volume = {7},

pages = {71\textendash95},

abstract = {Agentic AI systems, built upon large language models (LLMs) and deployed in multi-agent configurations, are redefining intelligence, autonomy, collaboration, and decision-making across enterprise and societal domains. This review presents a structured analysis of Trust, Risk, and Security Management (TRiSM) in the context of LLM-based Agentic Multi-Agent Systems (AMAS). We begin by examining the conceptual foundations of Agentic AI and highlight its architectural distinctions from traditional AI agents. We then adapt and extend the AI TRiSM framework for Agentic AI, structured around key pillars: Explainability, ModelOps, Security, Privacy and their Lifecycle Governance, each contextualized to the challenges of AMAS. A risk taxonomy is proposed to capture the unique threats and vulnerabilities of Agentic AI, ranging from coordination failures to prompt-based adversarial manipulation. To make coordination and tool use measurable in practice, we propose two metrics: the Component Synergy Score (CSS), which captures inter-agent enablement, and the Tool Utilization Efficacy (TUE), which evaluates whether tools are invoked correctly and efficiently. We further discuss strategies for improving explainability in Agentic AI, as well as approaches to enhancing security and privacy through encryption, adversarial robustness, and regulatory compliance. The review concludes with a research roadmap for the responsible development and deployment of Agentic AI, highlighting key directions to align emerging systems with TRiSM principles-ensuring safety, transparency, and accountability in their operation.},

keywords = {Adversarial robustness, Agentic AI, AI agents, AI governance, AI safety, Application security, Explainability, Human-in-the-Loop, LLM-based multi-agent systems, Model Privacy, ModelOps, Privacy-preserving AI, Risk management, TRiSM, Trustworthy AI},

pubstate = {published},

tppubtype = {article}

}

@article{sapkota_ai_2026,

title = {AI Agents vs. Agentic AI: A Conceptual taxonomy, applications and challenges},

author = {Ranjan Sapkota and Konstantinos I. Roumeliotis and Manoj Karkee},

url = {https://www.sciencedirect.com/science/article/pii/S1566253525006712},

doi = {https://doi.org/10.1016/j.inffus.2025.103599},

issn = {1566-2535},

year  = {2026},

date = {2026-01-01},

urldate = {2026-01-01},

journal = {Information Fusion},

volume = {126},

pages = {103599},

abstract = {Information fusion, in the context of the Generative AI era, must distinguish AI Agents from Agentic AI. This review critically distinguishes between AI Agents and Agentic AI, offering a structured, conceptual taxonomy, application mapping, and analysis of opportunities and challenges to clarify their divergent design philosophies and capabilities. We begin by outlining the search strategy and foundational definitions, characterizing AI Agents as modular systems driven and enabled by LLMs and LIMs for task-specific automation. Generative AI is positioned as a precursor providing the foundation, with AI agents advancing through tool integration, prompt engineering, and reasoning enhancements. We then characterize Agentic AI systems, which, in contrast to AI Agents, represent a paradigm shift marked by multi-agent collaboration, dynamic task decomposition, persistent memory, and coordinated autonomy. Through a chronological evaluation of architectural evolution, operational mechanisms, interaction styles, and autonomy levels, we present a comparative analysis across both AI agents and agentic AI paradigms. Application domains enabled by AI Agents such as customer support, scheduling, and data summarization are then contrasted with Agentic AI deployments in research automation, robotic coordination, and medical decision support. We further examine unique challenges in each paradigm including hallucination, brittleness, emergent behavior, and coordination failure, and propose targeted solutions such as ReAct loops, retrieval-augmented generation (RAG), automation coordination layers, and causal modeling. This work aims to provide a roadmap for developing robust, scalable, and explainable AI-driven systems.},

keywords = {Agentic AI, AI agents, Conceptual taxonomy, Context awareness, Multi-agent systems},

pubstate = {published},

tppubtype = {article}

}

@article{sapkota_comprehensive_2026,

title = {Comprehensive performance evaluation of YOLOv12, YOLO11, YOLOv10, YOLOv9 and YOLOv8 on detecting and counting fruitlet in complex orchard environments},

author = {Ranjan Sapkota and Zhichao Meng and Martin Churuvija and Xiaoqiang Du and Zenghong Ma and Manoj Karkee},

url = {https://www.sciencedirect.com/science/article/pii/S2949798126000050},

doi = {https://doi.org/10.1016/j.agrcom.2026.100125},

issn = {2949-7981},

year  = {2026},

date = {2026-01-01},

urldate = {2026-01-01},

journal = {Agriculture Communications},

volume = {4},

number = {1},

pages = {100125},

abstract = {This study systematically conducted an extensive real-world evaluation of all configurations of You Only Look Once (YOLO)-based object detection algorithms, including YOLOv8, YOLOv9, YOLOv10, YOLO11, and YOLOv12. Models were assessed using precision, recall, mean Average Precision at 50 % Intersection over Union (mAP@50), and computational efficiency across pre-processing, inference, and post-processing stages for detecting immature green fruitlets in commercial orchards. Field-level fruitlet counting was also validated using images captured with both Intel RealSense and iPhone 14 Pro Max sensors. YOLOv12l achieved the highest recall (0.900), while YOLOv10x and YOLOv9 GELAN-c reported the top precision scores of 0.908 and 0.903, respectively. YOLOv9 GELAN-base and GELAN-e achieved the highest mAP@50 (0.935), followed by YOLO11s (0.933) and YOLOv12l (0.931). In counting validation, YOLO11n demonstrated superior accuracy, with RMSE values of 4.51\textendash4.96 and MAE values of 3.85\textendash7.73 across four apple varieties. Sensor-specific training on Intel RealSense further improved detection performance. YOLO11n also recorded the fastest inference speed (2.4 ms), outperforming YOLOv8n, YOLOv9 GELAN-s, YOLOv10n, and YOLOv12n, affirming its suitability for real-time orchard applications.},

keywords = {agricultural automation, Fruitlet detection, Object detection, YOLO comparison, You Only Look Once (YOLO)},

pubstate = {published},

tppubtype = {article}

}

@article{roumeliotis_agentic_2026,

title = {Agentic AI With Orchestrator-Agent Trust: A Modular Visual Classification Framework With Trust-Aware Orchestration and RAG-Based Reasoning},

author = {Konstantinos I. Roumeliotis and Ranjan Sapkota and Manoj Karkee and Nikolaos D. Tselikas},

url = {https://ieeexplore.ieee.org/document/11373381/},

doi = {10.1109/ACCESS.2026.3662282},

issn = {2169-3536},

year  = {2026},

date = {2026-01-01},

urldate = {2026-01-01},

journal = {IEEE Access},

volume = {14},

pages = {26965\textendash26982},

abstract = {Modern Artificial Intelligence (AI) increasingly relies on multi-agent architectures that blend visual and language understanding. Yet, a pressing challenge remains: How can we trust these agents especially in zero-shot settings with no fine-tuning? We introduce a novel modular Agentic AI visual classification framework that integrates generalist multimodal agents with a non-visual reasoning orchestrator and a Retrieval-Augmented Generation (RAG) module. Applied to apple leaf disease diagnosis, we benchmark three configurations: (I) zero-shot with confidence-based orchestration, (II) fine-tuned agents with improved performance, and (III) trust-calibrated orchestration enhanced by CLIP-based image retrieval and re-evaluation loops. Using confidence calibration metrics (ECE, OCR, CCC), the orchestrator modulates trust across agents. Our results demonstrate a 77.94% accuracy improvement in the zero-shot setting using trust-aware orchestration and RAG, achieving 85.63% overall. GPT-4o showed better calibration, while Qwen-2.5-VL displayed overconfidence. Furthermore, image-RAG grounded predictions with visually similar cases, enabling correction of agent overconfidence via iterative re-evaluation. The proposed system separates perception (vision agents) from meta-reasoning (orchestrator), enabling scalable and interpretable multi-agent AI. This blueprint illustrates how Agentic AI can deliver trustworthy, modular, and transparent reasoning, and is extensible to diagnostics, biology, and other trust-critical domains. In doing so, we highlight Agentic AI not just as an architecture but as a paradigm for building reliable multi-agent intelligence. All models, prompts, results, and system components including the complete software source code are openly released to support reproducibility, transparency, and community benchmarking at our Github page.},

keywords = {Accuracy, Adaptation models, Agentic AI, Artificial intelligence, Calibration, Cognition, Costs, orchestrator agent trust, Retrieval augmented generation, retrieval augmented reasoning, Training, trust orchestration, visual classification, Visualization},

pubstate = {published},

tppubtype = {article}

}

@article{singh_incorporating_2025,

title = {Incorporating relative humidity in precipitation phase partitioning reduces model bias for some snow and streamflow metrics across the Northwest US},

author = {Bhupinderjeet Singh and Mingliang Liu and John T. Abatzoglou and Jennifer C. Adam and Kirti Rajagopalan},

url = {https://www.sciencedirect.com/science/article/pii/S0022169425011710},

doi = {10.1016/j.jhydrol.2025.133833},

issn = {0022-1694},

year  = {2025},

date = {2025-12-01},

urldate = {2025-12-01},

journal = {Journal of Hydrology},

volume = {662},

pages = {133833},

abstract = {While the importance of bivariate precipitation phase partitioning\textemdashthat incorporates both surface air temperature and relative humidity\textemdashhas been established for accurately estimating rain versus snow, hydrology models often rely on a simpler approach that uses only surface-temperature. We evaluate model bias changes for a suite of snow and streamflow metrics between temperature-based rain-snow partitioning (T-RSP) and temperature-relative-humidity-based rain-snow partitioning (TRH-RSP). We used the VIC-CropSyst coupled crop-hydrology model across the Pacific Northwest US as a case study. We found that transition to the TRH-RSP method resulted in a better match between modeled and observed (a) peak snow water equivalent (SWE) magnitude and timing (∼50% reduction in mean absolute bias), (b) daily SWE in winter months (reduction of relative bias from −30% to −4%), and (c) snow-start dates (mean reduction in bias from 7 days to 0 days) for the majority of the observational snow telemetry stations considered. Depending on the metric, 75\textendash88% of stations showed improvements. Most improvements are in the mid elevation stations. We also find improvements in estimates of basin-level streamflow and the ratio of peak SWE over streamflow. Elevation, temperature exposure, and meteorological bias partly explain the variability in performance improvements across stations. We did see a degradation in bias for snow-off dates. This is likely because meteorological bias and the modeled snowmelt dynamics\textemdashboth of which cannot be resolved by changing the precipitation partitioning\textemdashbecome important in the shoulder months at the end of the cold season. Overall, biases in SWE due to precipitation phase partitioning account for a substantial portion of the overall SWE bias\textemdashat least as much as, if not more than known precipitation biases. Transitioning from T-RSP to TRH-RSP can help us better understand model behavior, improve model accuracies, and better support management decision support for water resources, and prioritize improvements in melt dynamics to improve timing simulations.},

keywords = {AI, Water},

pubstate = {published},

tppubtype = {article}

}

@article{amogi_impact_2025,

title = {Impact of summer heat and mitigation strategies on apple (Cosmic Crisp®) fruit color dynamics quantified using crop physiology sensing system},

author = {Basavaraj R. Amogi and Lav R. Khot and Bernardita V. Sallato},

url = {https://www.sciencedirect.com/science/article/pii/S2666154325005344},

doi = {10.1016/j.jafr.2025.102163},

issn = {2666-1543},

year  = {2025},

date = {2025-10-01},

urldate = {2025-10-01},

journal = {Journal of Agriculture and Food Research},

volume = {23},

pages = {102163},

abstract = {Frequent summer heat waves significantly challenge global fruit production, including apples (Malus domestica Borkh.) grown in Washington State, USA. While growers employ heat mitigation strategies like evaporative cooling with overhead sprinklers, foggers, shade/drape netting, and protective sprays, these techniques can inadvertently compromise fruit coloration, a key quality attribute influencing harvest timing, marketability, and consumer acceptance. Thus, this study investigated whether continuous, in-orchard monitoring of fruit color and microclimatic conditions could help optimize mitigation practices without compromising fruit quality. Using a Crop Physiology Sensing System (CPSS), apple (Cosmic Crisp®) fruit color progression and ambient weather conditions were monitored at 5-min intervals throughout the 2022 growing season under fogging, netting, and untreated control treatments. CPSS with integrated RGB imaging data were contrasted with ambient air temperature (Tair) within each treatment using a custom developed algorithm. The algorithm allowed automated and daily quantification of fruit color metrics, including hue angle (h°), color transition from green to red (a∗), and chroma. Results suggest that prolonged daytime Tair exceeding 35 °C could cause significant degradation of red pigmentation (increasing h° and declining a∗). Netting caused overnight heat retention and delayed color recovery, whereas fogging effectively moderated the microclimate, preserving red coloration. Crucially, a nighttime drop in Tair to approximately 12 °C facilitated the reappearance of red coloration. To our knowledge, this is the first study to document both the degradation and subsequent reappearance of apple fruit coloration under field conditions. These findings suggest that continuous apple fruit color and ambient air temperature monitoring could be useful to effectively employ heat mitigation techniques, thereby improving fruit quality and market value at harvest.},

keywords = {AI, Farm Ops},

pubstate = {published},

tppubtype = {article}

}

@article{savalkar_earlier_2024,

title = {Earlier planting fails to replicate historical production conditions for US spring wheat under future climates},

author = {Supriya Savalkar and Michael Pumphrey and Kimberly Campbell and Fabio Scarpare and Tanvir Ferdousi and Samarth Swarup and Claudio St\"{o}ckle and Kirti Rajagopalan},

url = {https://www.nature.com/articles/s43247-025-02716-0},

doi = {10.1038/s43247-025-02716-0},

isbn = {2662-4435},

year  = {2025},

date = {2025-08-27},

urldate = {2024-10-01},

journal = {communications earth \& environment},

volume = {6},

issue = {1},

pages = {708},

publisher = {Research Square},

abstract = {Global warming and heat stress can adversely affect crop yields and quality. Earlier planting that shifts the growing season to cooler periods is a widely considered adaptation strategy in climate change literature. We ask: How effective is earlier planting in reducing high-temperature-exposure across growth stages? What are the associated temperature-exposure tradeoffs, and can historical conditions be matched? With US spring wheat as a case study, growth-stage-specific temperature exposure signatures are developed to estimate tradeoffs from earlier planting. While earlier planting does reduce exposure to critical and lethal high temperatures during reproductive stages, it fails to replicate historical production conditions. The Pacific Northwest is an exception, although tail-end growth stages may require management. Historically-equivalent planting windows narrow presenting logistical challenges. Therefore, while many climate-change assessments list earlier planting as an effective adaptation strategy, it may not be as effective when tradeoffs are considered, and consideration of other strategies will be important.},

note = {ISSN: 2693-5015},

keywords = {},

pubstate = {published},

tppubtype = {article}

}