2026
11.
William Solow; Paola Pesantez-Cabrera; Markus Keller; Lav Khot; Sandhya Saisubramanian; Alan Fern
A Hybrid Modeling Framework for Crop Prediction Tasks via Dynamic Parameter Calibration and Multi-Task Learning Working paper
2026, (arXiv:2603.15411 [cs.AI]).
Abstract | Links | BibTeX | Tags: Computer Science - Artificial Intelligence, Computer Science - Machine Learning
@workingpaper{solow_hybrid_2026,
title = {A Hybrid Modeling Framework for Crop Prediction Tasks via Dynamic Parameter Calibration and Multi-Task Learning},
author = {William Solow and Paola Pesantez-Cabrera and Markus Keller and Lav Khot and Sandhya Saisubramanian and Alan Fern},
url = {http://arxiv.org/abs/2603.15411},
doi = {10.48550/arXiv.2603.15411},
year = {2026},
date = {2026-05-01},
urldate = {2026-05-01},
publisher = {arXiv},
abstract = {Accurate prediction of crop states (e.g., phenology stages and cold hardiness) is essential for timely farm management decisions such as irrigation, fertilization, and canopy management to optimize crop yield and quality. While traditional biophysical models can be used for season-long predictions, they lack the precision required for site-specific management. Deep learning methods are a compelling alternative, but can produce biologically unrealistic predictions and require large-scale data. We propose a textbackslashemphhybrid modeling approach that uses a neural network to parameterize a differentiable biophysical model and leverages multi-task learning for efficient data sharing across crop cultivars in data limited settings. By predicting the textbackslashemphparameters of the biophysical model, our approach improves the prediction accuracy while preserving biological realism. Empirical evaluation using real-world and synthetic datasets demonstrates that our method improves prediction accuracy by 60textbackslash% for phenology and 40textbackslash% for cold hardiness compared to deployed biophysical models.},
note = {arXiv:2603.15411 [cs.AI]},
keywords = {Computer Science - Artificial Intelligence, Computer Science - Machine Learning},
pubstate = {published},
tppubtype = {workingpaper}
}
Accurate prediction of crop states (e.g., phenology stages and cold hardiness) is essential for timely farm management decisions such as irrigation, fertilization, and canopy management to optimize crop yield and quality. While traditional biophysical models can be used for season-long predictions, they lack the precision required for site-specific management. Deep learning methods are a compelling alternative, but can produce biologically unrealistic predictions and require large-scale data. We propose a textbackslashemphhybrid modeling approach that uses a neural network to parameterize a differentiable biophysical model and leverages multi-task learning for efficient data sharing across crop cultivars in data limited settings. By predicting the textbackslashemphparameters of the biophysical model, our approach improves the prediction accuracy while preserving biological realism. Empirical evaluation using real-world and synthetic datasets demonstrates that our method improves prediction accuracy by 60textbackslash% for phenology and 40textbackslash% for cold hardiness compared to deployed biophysical models.
12.
Rudrajit Choudhuri; Christopher A. Sanchez; Margaret Burnett; Anita Sarma
Thinking Less, Trusting More: GenAI’s Impacts on Students’ Cognitive Habits Miscellaneous
2026.
Abstract | Links | BibTeX | Tags:
@misc{choudhuri_thinking_2026,
title = {Thinking Less, Trusting More: GenAI’s Impacts on Students’ Cognitive Habits},
author = {Rudrajit Choudhuri and Christopher A. Sanchez and Margaret Burnett and Anita Sarma},
url = {https://papers.ssrn.com/abstract=6711809},
doi = {10.2139/ssrn.6711809},
year = {2026},
date = {2026-05-01},
urldate = {2026-05-01},
publisher = {Social Science Research Network},
address = {Rochester, NY},
abstract = {Context: Many students now use generative AI (genAI) in their coursework, yet its effects ontheir intellectual development remain poorly understood. While prior work has investigatedstudents’ cognitive offloading during episodic interactions, it remains unclear whether usinggenAI routinely is tied to more fundamental shifts in students’ thinking habits.Objectives: To explore this possibility, we investigate (RQ1-How): how students’ trust in androutine use of genAI affect their cognitive engagement\textemdashspecifically, reflection, the need forunderstanding, and critical thinking in STEM coursework. Further, we investigate (RQ2-Who):which students are particularly vulnerable to these cognitive disengagement effects.Methods: We drew on dual-process theory, cognitive offloading, and the automation biasliterature to develop a statistical model explaining how and to what extent students’ trust-driven routine use of genAI affected their cognitive engagement habits in STEM coursework, and how these effects differed across students’ diverse cognitive styles. We empirically evaluated this model using Partial Least Squares Structural Equation Modeling on survey data from 299 STEM students across five North American universities.Results: Students who trusted and routinely used genAI reported significantly lower cognitiveengagement. Unexpectedly, students with higher technophilic motivations, risk tolerance, andcomputer self-efficacy\textemdashtraits often celebrated in STEM\textemdashwere more prone to these effects.Interestingly, students’ prior experience with genAI or academia did not protect them fromcognitively disengaging.Conclusion: Our findings suggest a potential cognitive debt cycle in which routine genAIuse progressively weakens students’ intellectual habits, potentially driving over-reliance andescalating usage.This poses critical challenges for curricula and genAI system design, requiring interventions that actively support cognitive engagement.},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Context: Many students now use generative AI (genAI) in their coursework, yet its effects ontheir intellectual development remain poorly understood. While prior work has investigatedstudents’ cognitive offloading during episodic interactions, it remains unclear whether usinggenAI routinely is tied to more fundamental shifts in students’ thinking habits.Objectives: To explore this possibility, we investigate (RQ1-How): how students’ trust in androutine use of genAI affect their cognitive engagement—specifically, reflection, the need forunderstanding, and critical thinking in STEM coursework. Further, we investigate (RQ2-Who):which students are particularly vulnerable to these cognitive disengagement effects.Methods: We drew on dual-process theory, cognitive offloading, and the automation biasliterature to develop a statistical model explaining how and to what extent students’ trust-driven routine use of genAI affected their cognitive engagement habits in STEM coursework, and how these effects differed across students’ diverse cognitive styles. We empirically evaluated this model using Partial Least Squares Structural Equation Modeling on survey data from 299 STEM students across five North American universities.Results: Students who trusted and routinely used genAI reported significantly lower cognitiveengagement. Unexpectedly, students with higher technophilic motivations, risk tolerance, andcomputer self-efficacy—traits often celebrated in STEM—were more prone to these effects.Interestingly, students’ prior experience with genAI or academia did not protect them fromcognitively disengaging.Conclusion: Our findings suggest a potential cognitive debt cycle in which routine genAIuse progressively weakens students’ intellectual habits, potentially driving over-reliance andescalating usage.This poses critical challenges for curricula and genAI system design, requiring interventions that actively support cognitive engagement.
13.
Viticulture & Enology Washington State University (Ed.)
Viticulture & Enology Extension News – Spring 2026 Online
& Enology Washington State University, Viticulture (Ed.): 2026, visited: 13.04.2026.
@online{noauthor_viticulture_nodate,
title = {Viticulture \& Enology Extension News \textendash Spring 2026},
editor = {Viticulture \& Enology Washington State University},
url = {https://wine.wsu.edu/viticulture-enology-extension-news-spring-2026/#CropAI},
year = {2026},
date = {2026-04-13},
urldate = {2026-04-13},
keywords = {},
pubstate = {published},
tppubtype = {online}
}
14.
Keyvan Malek; Matthew Yourek; Jennifer Adam; Alan Hamlet; Kirti Rajagopalan; Patrick Reed
RColSim: An R-Based Open-Source Regional Water Management Model for the Columbia River Basin Journal Article
In: Journal of Open Research Software, vol. 14, no. 1, 2026, ISSN: 2049-9647.
Abstract | Links | BibTeX | Tags:
@article{malek_rcolsim_2026,
title = {RColSim: An R-Based Open-Source Regional Water Management Model for the Columbia River Basin},
author = {Keyvan Malek and Matthew Yourek and Jennifer Adam and Alan Hamlet and Kirti Rajagopalan and Patrick Reed},
url = {https://openresearchsoftware.metajnl.com/articles/10.5334/jors.406},
doi = {10.5334/jors.406},
issn = {2049-9647},
year = {2026},
date = {2026-04-01},
urldate = {2026-04-01},
journal = {Journal of Open Research Software},
volume = {14},
number = {1},
abstract = {We present RColSim, a script-based water management model that simulates the operation of reservoir systems of the Columbia River Basin located in the Pacific Northwest. The model simulates the operation of 46 major storage and run-of-the-river dams on the Columbia River, given rule curves and flow targets for system-wide flood control, irrigation, environmental, and hydroelectricity generation. RColSim is written in the R programming language and is publicly available on GitHub and Zenodo. The model can be downloaded, modified, and reused in accordance with the MIT open-source license terms of use. The model provides an open-source simulation tool for researchers, practitioners, and policymakers to investigate the ramifications of various planning and operation scenarios and stressors for Columbia River Basin stakeholders.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
We present RColSim, a script-based water management model that simulates the operation of reservoir systems of the Columbia River Basin located in the Pacific Northwest. The model simulates the operation of 46 major storage and run-of-the-river dams on the Columbia River, given rule curves and flow targets for system-wide flood control, irrigation, environmental, and hydroelectricity generation. RColSim is written in the R programming language and is publicly available on GitHub and Zenodo. The model can be downloaded, modified, and reused in accordance with the MIT open-source license terms of use. The model provides an open-source simulation tool for researchers, practitioners, and policymakers to investigate the ramifications of various planning and operation scenarios and stressors for Columbia River Basin stakeholders.
15.
Nibir Chandra Mandal; Oishee Bintey Hoque; Mandy Wilson; Samarth Swarup; Sayjro Nouwakpo; Abhijin Adiga; Madhav Marathe
IRRISIGHT: A Large-Scale Multimodal Dataset and Scalable Pipeline to Address Irrigation and Water Management in Agriculture Journal Article
In: Advances in Neural Information Processing Systems, vol. 38, 2026.
@article{mandal_irrisight_2026,
title = {IRRISIGHT: A Large-Scale Multimodal Dataset and Scalable Pipeline to Address Irrigation and Water Management in Agriculture},
author = {Nibir Chandra Mandal and Oishee Bintey Hoque and Mandy Wilson and Samarth Swarup and Sayjro Nouwakpo and Abhijin Adiga and Madhav Marathe},
url = {https://proceedings.neurips.cc/paper_files/paper/2025/hash/0821a1413339bf79ba01876783d95c53-Abstract-Datasets_and_Benchmarks_Track.html},
year = {2026},
date = {2026-04-01},
urldate = {2026-04-01},
journal = {Advances in Neural Information Processing Systems},
volume = {38},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
16.
Deanna Flynn; Abhinav Jain; Heather Knight; Cristina G. Wilson; Cindy Grimm
Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives Journal Article
In: 2026.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{flynn_uncovering_2026,
title = {Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives},
author = {Deanna Flynn and Abhinav Jain and Heather Knight and Cristina G. Wilson and Cindy Grimm},
url = {https://journals.ashs.org/view/journals/horttech/36/2/article-p325.xml},
doi = {10.21273/HORTTECH05817-25},
year = {2026},
date = {2026-04-01},
urldate = {2026-04-01},
chapter = {HortTechnology},
abstract = {Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives
17.
Alec Busteed; Jimena Noa-Guevara; Lais Alexandra Castro; Dahana Moz Ruiz; Sadia Afroz; Iman Mokraoui; Prisha Velhal; Patricia Morreale; Anita Sarma; Margaret Burnett
“Fast, easy, simple”? SES-diverse transfer students' sociotechnical experiences registering for classes Proceedings Article
In: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems, pp. 1–23, Association for Computing Machinery, New York, NY, USA, 2026, ISBN: 979-8-4007-2278-3.
Abstract | Links | BibTeX | Tags: AI, Humans
@inproceedings{busteed_fast_2026,
title = {“Fast, easy, simple”? SES-diverse transfer students\' sociotechnical experiences registering for classes},
author = {Alec Busteed and Jimena Noa-Guevara and Lais Alexandra Castro and Dahana Moz Ruiz and Sadia Afroz and Iman Mokraoui and Prisha Velhal and Patricia Morreale and Anita Sarma and Margaret Burnett},
url = {https://dl.acm.org/doi/10.1145/3772318.3791127},
doi = {10.1145/3772318.3791127},
isbn = {979-8-4007-2278-3},
year = {2026},
date = {2026-04-01},
urldate = {2026-04-01},
booktitle = {Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems},
pages = {1\textendash23},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {CHI \'26},
abstract = {Recruiting, retaining, and educating students in computing is a frequent research topic in CHI. However, students’ sociotechnical experiences of registering for classes are understudied\textemdashespecially those of socioeconomic-diverse students. These experiences matter: research shows that registration problems bring long-term consequences to student successes. We investigate students’ socioeconomic status (SES) impact on registration experiences through three studies: a case study with education professionals using an emerging analytic method, SocioeconomicMag (SESMag); interviews with faculty/staff/students from 8 universities; and observations of 14 SES-diverse students registering for classes. Results showed: (1) 5 SES-inclusivity bugs which arose 30 times, 72% more often by lower-SES students than by higher-SES students. (2) 6/7 lower-SES students (but only 2/7 higher-SES students) expected downstream problems from the registration issues. (3) The risk-to-negative-outcomes rate was 3 times higher for lower-SES students. (4) The issues generalized across 8 universities and potentially to \>700 other universities who use the same registration portal.},
keywords = {AI, Humans},
pubstate = {published},
tppubtype = {inproceedings}
}
Recruiting, retaining, and educating students in computing is a frequent research topic in CHI. However, students’ sociotechnical experiences of registering for classes are understudied—especially those of socioeconomic-diverse students. These experiences matter: research shows that registration problems bring long-term consequences to student successes. We investigate students’ socioeconomic status (SES) impact on registration experiences through three studies: a case study with education professionals using an emerging analytic method, SocioeconomicMag (SESMag); interviews with faculty/staff/students from 8 universities; and observations of 14 SES-diverse students registering for classes. Results showed: (1) 5 SES-inclusivity bugs which arose 30 times, 72% more often by lower-SES students than by higher-SES students. (2) 6/7 lower-SES students (but only 2/7 higher-SES students) expected downstream problems from the registration issues. (3) The risk-to-negative-outcomes rate was 3 times higher for lower-SES students. (4) The issues generalized across 8 universities and potentially to >700 other universities who use the same registration portal.
18.
Prathyush Sambaturu; Manisha Sudhir; Hongze Chen; Anil Vullikanti; Rangaswamy Muniappan; Abhijin Adiga
Scenario-aware control of multipathway spread processes: Application to biological invasions Journal Article
In: PNAS Nexus, vol. 5, no. 3, pp. pgag036, 2026, ISSN: 2752-6542.
Abstract | Links | BibTeX | Tags: AI, Water
@article{sambaturu_scenario-aware_2026,
title = {Scenario-aware control of multipathway spread processes: Application to biological invasions},
author = {Prathyush Sambaturu and Manisha Sudhir and Hongze Chen and Anil Vullikanti and Rangaswamy Muniappan and Abhijin Adiga},
url = {https://doi.org/10.1093/pnasnexus/pgag036},
doi = {10.1093/pnasnexus/pgag036},
issn = {2752-6542},
year = {2026},
date = {2026-03-01},
urldate = {2026-03-01},
journal = {PNAS Nexus},
volume = {5},
number = {3},
pages = {pgag036},
abstract = {Optimal control of spread processes over networks is a challenging problem, even for simple diffusion models. Real-world processes\textemdashsuch as infectious disease outbreaks and biological invasions\textemdashoften involve multiple spread pathways and time-varying network dynamics. In this work, we address the problem of region-wide interventions, where the goal is to select an optimal set of regions (groups of nodes) in a network to minimize spread, subject to budget constraints, intervention delays, and a given spread scenario which reflects prior knowledge of the process\textemdashsuch as initial infection locations, parameter estimates, and other context-specific assumptions. We present a general approach based on integer linear programming and sample average approximation, applicable across a broad class of diffusion models. We also establish theoretical performance guarantees for our method within the bicriteria approximation framework. To demonstrate its effectiveness, we apply the approach to model the spread of a representative agricultural pest. Our method yields near-optimal solutions and consistently outperforms standard baselines. The results emphasize the value of scenario-specific intervention strategies, showing that early action can significantly reduce spread under limited budgets and produce stable outcomes even under model uncertainty.},
keywords = {AI, Water},
pubstate = {published},
tppubtype = {article}
}
Optimal control of spread processes over networks is a challenging problem, even for simple diffusion models. Real-world processes—such as infectious disease outbreaks and biological invasions—often involve multiple spread pathways and time-varying network dynamics. In this work, we address the problem of region-wide interventions, where the goal is to select an optimal set of regions (groups of nodes) in a network to minimize spread, subject to budget constraints, intervention delays, and a given spread scenario which reflects prior knowledge of the process—such as initial infection locations, parameter estimates, and other context-specific assumptions. We present a general approach based on integer linear programming and sample average approximation, applicable across a broad class of diffusion models. We also establish theoretical performance guarantees for our method within the bicriteria approximation framework. To demonstrate its effectiveness, we apply the approach to model the spread of a representative agricultural pest. Our method yields near-optimal solutions and consistently outperforms standard baselines. The results emphasize the value of scenario-specific intervention strategies, showing that early action can significantly reduce spread under limited budgets and produce stable outcomes even under model uncertainty.
19.
Krishu K. Thapa; Supriya Savalkar; Bhupinderjeet Singh; Trong Nghia Hoang; Kirti Rajagopalan; Ananth Kalyanaraman
ForeSWE: Forecasting Snow-Water Equivalent with an Uncertainty-Aware Attention Model Journal Article
In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 40, no. 46, pp. 39348–39357, 2026, ISSN: 2374-3468.
Abstract | Links | BibTeX | Tags:
@article{thapa_foreswe_2026,
title = {ForeSWE: Forecasting Snow-Water Equivalent with an Uncertainty-Aware Attention Model},
author = {Krishu K. Thapa and Supriya Savalkar and Bhupinderjeet Singh and Trong Nghia Hoang and Kirti Rajagopalan and Ananth Kalyanaraman},
url = {https://ojs.aaai.org/index.php/AAAI/article/view/41284},
doi = {10.1609/aaai.v40i46.41284},
issn = {2374-3468},
year = {2026},
date = {2026-03-01},
urldate = {2026-03-01},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
volume = {40},
number = {46},
pages = {39348\textendash39357},
abstract = {Various complex water management decisions are made in snow-dominant watersheds with the knowledge of Snow-Water Equivalent (SWE)\textemdasha key measure widely used to estimate the water content of a snowpack. However, forecasting SWE is challenging because SWE is influenced by various factors including topography and an array of environmental conditions, and has therefore been observed to be spatio-temporally variable. Classical approaches to SWE forecasting have not adequately utilized these spatial/temporal correlations, nor do they provide uncertainty estimates \textemdash which can be of significant value to the decision maker. In this paper, we present ForeSWE, a new probabilistic spatio-temporal forecasting model that integrates deep learning and classical probabilistic techniques. The resulting model features a combination of an attention mechanism to integrate spatiotemporal features and interactions, alongside a Gaussian process module that provides principled quantification of prediction uncertainty. We evaluate the model on data from 512 Snow Telemetry (SNOTEL) stations in the Western US. The results show significant improvements in both forecasting accuracy and prediction interval compared to state-of-the-art approaches. The results also serve to highlight the efficacy in uncertainty estimates between different approaches. Collectively, these findings have provided a platform for deployment and feedback by the water management community.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Various complex water management decisions are made in snow-dominant watersheds with the knowledge of Snow-Water Equivalent (SWE)—a key measure widely used to estimate the water content of a snowpack. However, forecasting SWE is challenging because SWE is influenced by various factors including topography and an array of environmental conditions, and has therefore been observed to be spatio-temporally variable. Classical approaches to SWE forecasting have not adequately utilized these spatial/temporal correlations, nor do they provide uncertainty estimates — which can be of significant value to the decision maker. In this paper, we present ForeSWE, a new probabilistic spatio-temporal forecasting model that integrates deep learning and classical probabilistic techniques. The resulting model features a combination of an attention mechanism to integrate spatiotemporal features and interactions, alongside a Gaussian process module that provides principled quantification of prediction uncertainty. We evaluate the model on data from 512 Snow Telemetry (SNOTEL) stations in the Western US. The results show significant improvements in both forecasting accuracy and prediction interval compared to state-of-the-art approaches. The results also serve to highlight the efficacy in uncertainty estimates between different approaches. Collectively, these findings have provided a platform for deployment and feedback by the water management community.
20.
Oishee Bintey Hoque; Nibir Chandra Mandal; Kyle Luong; Amanda Wilson; Samarth Swarup; Madhav Marathe; Abhijin Adiga
PRISM-CAFO: Prior-conditioned Remote-sensing Infrastructure Segmentation and Mapping for CAFOs Proceedings Article
In: 2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pp. 2083–2093, 2026, ISSN: 2642-9381, (ISSN: 2642-9381).
Abstract | Links | BibTeX | Tags: Antennas, Circuits and systems, classification, Earth Observing System, Feeds, Filtering, Filters, infrastructure-guided vision, interpretibility, Location awareness, masked attention pooling, multimodal feature fusion, objecj detection, Pixel, Remote Sensing, Satellite images, Segmentation, Sentinel-2, spatial reasoning, synthetic infrastructure mask
@inproceedings{hoque_prism-cafo_2026,
title = {PRISM-CAFO: Prior-conditioned Remote-sensing Infrastructure Segmentation and Mapping for CAFOs},
author = {Oishee Bintey Hoque and Nibir Chandra Mandal and Kyle Luong and Amanda Wilson and Samarth Swarup and Madhav Marathe and Abhijin Adiga},
url = {https://ieeexplore.ieee.org/abstract/document/11492693},
doi = {10.1109/WACV61042.2026.00207},
issn = {2642-9381},
year = {2026},
date = {2026-03-01},
urldate = {2026-03-01},
booktitle = {2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
pages = {2083\textendash2093},
abstract = {Large-scale livestock operations pose significant risks to human health and the environment, while also being vulnerable to threats such as infectious diseases and extreme weather events. As the number of such operations continues to grow, accurate and scalable mapping has become increasingly important. In this work, we present an infrastructure-first, explainable pipeline for identifying and characterizing Concentrated Animal Feeding Operations (CAFOs) from aerial and satellite imagery. Our method (i) detects candidate infrastructure (e.g., barns, feedlots, manure lagoons, silos) with a domain-tuned YOLOv8 detector, then derives SAM2 masks from these boxes and filters component-specific criteria; (ii) extracts structured descriptors (e.g., counts, areas, orientations, and spatial relations) and fuses them with deep visual features using a lightweight spatial cross-attention classifier; and (iii) outputs both CAFO type predictions and mask-level attributions that link decisions to visible infrastructure. Through comprehensive evaluation, we show that our approach achieves state-of-the-art performance, with Swin-B+PRISM-CAFO surpassing the best performing baseline by up to 15%. Beyond strong predictive performance across diverse U.S. regions, we run systematic gradient\textendashactivation analyses that quantify the impact of domain priors and show how specific infrastructure (e.g., barns, lagoons) shapes classification decisions. We release code, infrastructure masks, and descriptors to support transparent, scalable monitoring of livestock infrastructure, enabling risk modeling, change detection, and targeted regulatory action. Github: https://github.com/Nibir088/PRISM-CAFO.},
note = {ISSN: 2642-9381},
keywords = {Antennas, Circuits and systems, classification, Earth Observing System, Feeds, Filtering, Filters, infrastructure-guided vision, interpretibility, Location awareness, masked attention pooling, multimodal feature fusion, objecj detection, Pixel, Remote Sensing, Satellite images, Segmentation, Sentinel-2, spatial reasoning, synthetic infrastructure mask},
pubstate = {published},
tppubtype = {inproceedings}
}
Large-scale livestock operations pose significant risks to human health and the environment, while also being vulnerable to threats such as infectious diseases and extreme weather events. As the number of such operations continues to grow, accurate and scalable mapping has become increasingly important. In this work, we present an infrastructure-first, explainable pipeline for identifying and characterizing Concentrated Animal Feeding Operations (CAFOs) from aerial and satellite imagery. Our method (i) detects candidate infrastructure (e.g., barns, feedlots, manure lagoons, silos) with a domain-tuned YOLOv8 detector, then derives SAM2 masks from these boxes and filters component-specific criteria; (ii) extracts structured descriptors (e.g., counts, areas, orientations, and spatial relations) and fuses them with deep visual features using a lightweight spatial cross-attention classifier; and (iii) outputs both CAFO type predictions and mask-level attributions that link decisions to visible infrastructure. Through comprehensive evaluation, we show that our approach achieves state-of-the-art performance, with Swin-B+PRISM-CAFO surpassing the best performing baseline by up to 15%. Beyond strong predictive performance across diverse U.S. regions, we run systematic gradient–activation analyses that quantify the impact of domain priors and show how specific infrastructure (e.g., barns, lagoons) shapes classification decisions. We release code, infrastructure masks, and descriptors to support transparent, scalable monitoring of livestock infrastructure, enabling risk modeling, change detection, and targeted regulatory action. Github: https://github.com/Nibir088/PRISM-CAFO.
