AI-DRIVEN APPROACHES TO CROP YIELD PREDICTION AND MANAGEMENT: TOWARDS SUSTAINABLE AND SMART AGRICULTURE"
Fuente:
Zenodo
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Recurso digital |
| Lingua: | inglese |
| Pubblicazione: |
Zenodo
2025
|
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866901099246518272 |
|---|---|
| author | Asst.Prof. Kruthika.R SUDHAKARAN T |
| author_facet | Asst.Prof. Kruthika.R SUDHAKARAN T |
| contents | <h1><strong><span lang="EN-IN">Abstract<span> </span></span></strong></h1> <p><strong><em><span lang="EN-IN">Climate change, soil degradation, and pest outbreaks are increasingly threatening agricultural productivity, heightening the need for accurate crop-yield forecasting. Artificial intelligence (AI) – leveraging satellite imagery, sensor networks, and farm data – offers powerful new tools to predict yields with high precision. This paper synthesizes current challenges and solutions in AI-driven yield prediction. We review how machine learning (ML) and deep learning models (e.g. Random Forests, gradient boosting, convolutional neural networks, recurrent networks) have been applied to large multi-modal datasets (satellite VIs, weather, soil sensors, farm records) to forecast yields . Through case studies (e.g. winter wheat in China, sugarcane in Brazil, mixed crops in Ethiopia), we illustrate how AI models (LSTM, RF, GA regression) can achieve high accuracy (e.g. R² up to 0.92 or RMSE reductions) by integrating time-series and spatial data. We identify critical barriers, including limited labelled data, rural connectivity and infrastructure gaps, and the opaque “black-box” nature of AI models. We propose solutions such as federated learning for privacy-preserving multi-farm training, edge computing to enable on-device analytics, transfer learning to adapt models across regions, and explainable AI (XAI) to build trust. We also discuss future directions like climate-resilient modeling, farm </span></em></strong><strong><em><span lang="EN-IN">digital twins</span></em></strong><strong><em><span lang="EN-IN">, and inclusive tools for smallholder farmers. The paper concludes with recommendations to bridge AI research and practical deployment, emphasizing scalable data strategies and user-friendly decision-support platforms.</span></em></strong></p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18065506 |
| institution | Zenodo |
| language | eng |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | AI-DRIVEN APPROACHES TO CROP YIELD PREDICTION AND MANAGEMENT: TOWARDS SUSTAINABLE AND SMART AGRICULTURE" Asst.Prof. Kruthika.R SUDHAKARAN T <h1><strong><span lang="EN-IN">Abstract<span> </span></span></strong></h1> <p><strong><em><span lang="EN-IN">Climate change, soil degradation, and pest outbreaks are increasingly threatening agricultural productivity, heightening the need for accurate crop-yield forecasting. Artificial intelligence (AI) – leveraging satellite imagery, sensor networks, and farm data – offers powerful new tools to predict yields with high precision. This paper synthesizes current challenges and solutions in AI-driven yield prediction. We review how machine learning (ML) and deep learning models (e.g. Random Forests, gradient boosting, convolutional neural networks, recurrent networks) have been applied to large multi-modal datasets (satellite VIs, weather, soil sensors, farm records) to forecast yields . Through case studies (e.g. winter wheat in China, sugarcane in Brazil, mixed crops in Ethiopia), we illustrate how AI models (LSTM, RF, GA regression) can achieve high accuracy (e.g. R² up to 0.92 or RMSE reductions) by integrating time-series and spatial data. We identify critical barriers, including limited labelled data, rural connectivity and infrastructure gaps, and the opaque “black-box” nature of AI models. We propose solutions such as federated learning for privacy-preserving multi-farm training, edge computing to enable on-device analytics, transfer learning to adapt models across regions, and explainable AI (XAI) to build trust. We also discuss future directions like climate-resilient modeling, farm </span></em></strong><strong><em><span lang="EN-IN">digital twins</span></em></strong><strong><em><span lang="EN-IN">, and inclusive tools for smallholder farmers. The paper concludes with recommendations to bridge AI research and practical deployment, emphasizing scalable data strategies and user-friendly decision-support platforms.</span></em></strong></p> |
| title | AI-DRIVEN APPROACHES TO CROP YIELD PREDICTION AND MANAGEMENT: TOWARDS SUSTAINABLE AND SMART AGRICULTURE" |
| url | https://doi.org/10.5281/zenodo.18065506 |