Smart Farming with IoT and Machine Learning for Crop Recommendation and Disease Detection

Abstract

Key challenges in traditional agriculture include subjective crop recommendation methods based on farmer expe- rience, inefficient plant disease detection techniques reliant on visual inspection, and rudimentary environmental monitoring methods using manual observations. These limitations hinder optimal crop management and environmental control, leading to reduced productivity and increased vulnerability to pests and diseases. Smart agricultural system addresses the limitations imposed by outdated farming practices by incorporating IoT sensors and Machine Learning (ML) algorithms to facilitate data-driven decision-making and optimize farming processes. Key functionalities include crop recommendation, plant disease prediction, soil moisture monitoring, and humidity and temperature monitoring. Crop recommendation is facilitated by ML algorithms, specifically Random Forest, which analyses collected data to suggest suitable crops for specific geographic areas. Disease prediction employs TensorFlow models to accurately detect and diagnose plant diseases based on image data. Soil moisture monitoring is achieved through soil sensor, providing real-time data on soil water content, while humidity and temperature levels are monitored using DHT11 sensor. These environmental parameters are crucial for maintaining optimal growing conditions and mitigating risks associated with climate variability. Through the integration of IoT and ML technologies, our system offers a practical solution to enhance agricultural practices in resource-constrained settings. By providing farmers with actionable insights and decision support, we aim to improve crop yields, optimize resource utilization, and promote sustain- able agriculture.

Introduction

1. Overview

Traditional agriculture faces major challenges in productivity and sustainability due to inefficient crop recommendation methods, poor disease detection, and outdated environmental monitoring. Crop yield is influenced by various biotic (pests, humans, microbes) and abiotic (climate, soil, temperature) factors, further complicated by climate change and excessive chemical use. There is a critical need for early disease prediction and smart environmental monitoring to improve outcomes.

2. Literature Survey

Several studies emphasize the potential of IoT and Machine Learning (ML) in modern agriculture:

• Farooq et al.: Focus on IoT-based greenhouse management systems, including security and communication challenges.

• Saleem et al.: Developed a deep learning model using IoT-enabled environmental data to predict pest outbreaks with 94% accuracy.

• Liu et al.: Created an ML model (using Multiple Linear Regression) for early disease prediction in tea plantations using real-time environmental data.

• Raja et al.: Proposed ML-based crop yield prediction by selecting relevant features such as humidity and rainfall.

• Zhou & Yin: Conducted a large-scale review on Digital Agriculture (DA), highlighting key research areas such as AI, IoT, remote sensing, and big data.

3. Proposed System

A Sustainable Agriculture Management System is developed integrating IoT and ML for:

• Crop recommendation

• Plant disease detection

• Temperature, humidity, and soil moisture monitoring

Methodology Steps:

1. Data Collection: Sensors like DHT11 and soil moisture sensors collect real-time data.

2. Communication: Uses serial communication to transfer sensor data.

3. Data Processing: Involves cleaning, mining, and feature engineering.

4. Analysis: Applies ML models (e.g., Random Forest, VGG16) for predictions.

5. Result Prediction: Generates actionable insights like suitable crops, moisture levels, and disease status.

6. End-Users: Farmers use a web interface to make informed decisions.

4. Components Used

• DHT11 Sensor: Measures temperature and humidity.

• Soil Sensors: Measure moisture and support precise irrigation.

• Arduino Uno: Central microcontroller connecting all components.

5. Implementation Details

• Crop Recommendation: Uses Random Forest trained on a Kaggle dataset with 2200 samples; achieved 99.2% accuracy.

• Disease Detection: Uses VGG16 neural network, reaching 97.8% accuracy on the Plant Village dataset.

• Environmental Monitoring: Categorizes temperature, humidity, and moisture into Low, Normal, Optimal, or High using real-time sensor data.

6. Results & Impact

Sensor Monitoring Results:

Disease Detection Metrics:

• Accuracy: 97.8%

• Precision: 97.4%

• Recall: 97.6%

• F1-Score: 97.5%

Crop Recommendation Model Comparison:

Real-World Impact:

Conclusion

In conclusion, the emergence of Smart Agriculture Management using IoT and Machine Learning signifies a significant advancement in revolutionizing the agricultural sector. Traditional farming methods often face challenges related to environmental monitoring, this system effectively addresses existing shortcomings in farming practices. The Random Forest algorithm achieved a classification accuracy of 99.2%, outperforming other machine learning models such as Decision Tree (95.6%), Support Vector Machine (96.8%), and K-Nearest Neighbours (94.3%). Precision and recall values remained consistently above 97%, indicating the robustness of the model in recommending suitable crops based on soil and environmental features. Feature importance analysis revealed that soil pH (22%), rainfall (18%), and nitrogen content (16%) were the most influential parameters in determining crop suitability. Ultimately, the adoption of Smart Agriculture Management not only enhances productivity and sustainability but also showcases the trans- formative potential of innovative technologies in modernizing farming methods.

References

[1] Muhammad Shoaib Farooq, Rizwan Javid,Shamyla Riaz and Zabihul- lah Atal, “IoT Based Smart Greenhouse Framework and Control for Sustainable Agriculture”, 2022 IEEE Volume 10 [2] Rana Muhammad Salim, Rab Nawaz Bashir, Muhammad Faheem, Mohd Anul Haq, Ahmed Alhussen, Zamil S. Alzamil, Shakir Khan,”Internet of Things based weekly crop pest prediction by using Deep Neural Network” 2023 IEEE Volume 11 [3] Ahmad Ali Alzubi and Kalda Galyna ,“Artificial Intelligence and Internet of Things for Sustainable Farming and Smart Agriculture ”,2023 IEEE Volume 11 [4] Zhiyan Liu, Rab Nawaz Bashir, Salman Iqbal, Malik Muhammad Ali Shahid, Muhammad Taushif, and Qasim Umer “Internet of Things (IoT) and Machine Learning Model of Plant Disease Prediction—Blister Blight for Tea Plant” 2022 IEEE Volume 10 [5] S.P Raja, Barbara Sawicka , Zoran Stamenkovic, and G. Mari- ammal,”Crop Prediction Based on Characteristics of the Agricultural Environment Using Various Feature Selection Techniques and Classi- fiers”,2022 IEEE Volume 10 [6] Sameer Qazi , Bilal A Khawaja and Qazi Umar Farooq, “IoT- Equipped and AI-Enabled Next Generation Smart Agriculture: A Critical Review [7] , Current Challenges and Future Trends ”, 2022 IEEE Volume 10 [8] R. Akhter and S. A. Sofi, “Precision agriculture using IoT data analytics 1524 and machine learning,” J. King Saud Univ.-Comput. Inf. Sci., vol. 34, 1525 no. 8, pp. 5602–5618, 2021 [9] Bhat S A and Huang N F, “Big data and AI revolution in precision agriculture: Survey and challenges”,2021 IEEE Access [10] Carlos Miskinis, ”How IoT can help solve the biggest problem in AG”, blog Jan 2019. [11] O. Elijah, T. A. Rahman, I. Orikumhi, C. Y. Leow, and M. N. Hindia, 1922 “An overview of Internet of Things (IoT) and data analytics in agriculture: Benefits and challenges,” 2018 IEEE , vol. 5, no. 5, 1924 [12] pp. 3758–3773 [13] R. Dagar, S. Som, and S. K. Khatri, “Smart farming-IoT in agricul- ture”,2018 ICIRCA [14] Rongji Zhou, Yuyan Yin,”Digital Agriculture : Mapping Knowledge Structure and Trends”,2017 IEEE

Copyright

Copyright © 2025 Rehna R S, Anjitha Joy, Apoorva R P, Devarenjini P, Devika Vijayan. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.