A Systematic Review of IoT and AI Technologies in Indian Agricultural Practices

a. Search criteria

Academic data sources, including IEEE Xplore, Web of Science, ScienceDirect, and Scopus, are used to retrieve and analyze the pertinent literature studies. The most recent studies being carried out in India’s agriculture sector are better understood thanks to these publications. Using exact search terms and keywords, the writers have carefully examined the relevant research articles to address the developed RQs. The following keywords were used to accomplish the study’s goals: “Internet of Things AND Precision Agriculture,” “Artificial Intelligence AND Precision Agriculture,” “Sustainable Agriculture in India AND Artificial Intelligence,” “Sustainable Agriculture in India AND Internet of Things,” and “Sustainable Agriculture in India AND Artificial Intelligence AND Internet of Things.” The research studies retrieved using the initial search keywords showed significant overlap. Therefore, several criteria are applied in the assessment, including highly cited publications, recent articles, sorting by date, and sorting by relevance.

b. Data collection

Only English-language journal articles, conference proceedings, and early-access articles are included in the original search. The Digital India Mission has created a demand for IoT and AI in Indian agriculture, which is still in its infancy. As a result, early access publications, as well as statistical data and information from government websites, have been incorporated into the evaluation to provide insightful information.

c. Inclusion and exclusion criteria

The next step is to screen the relevant papers after retrieving them from the appropriate databases using search parameters. Titles and abstracts are used to filter the documents, and duplicates or unnecessary copies are eliminated. The references of the documents are reviewed to identify any other important studies. Research studies that are deemed irrelevant or lack peer review are excluded.

Furthermore, research that is not related to the field of Indian agriculture is also excluded. It includes the studies that address the objectives of the research study. Consequently, 78 research studies are chosen for in-depth examination. The inclusion and exclusion criteria are explained in detail with the help of the PRISMA flow diagram in Figure 2.

Figure 2:

d. Quality assessment criteria

Table 3 demonstrates that the quality assessment criteria help to answer the crucial RQs that will substantially contribute to the review.

e. Data extraction

Qualitative analysis is conducted using the data from research publications and reports that met the study selection criteria. The content is categorized and analyzed using the thematic map to create the parts and subsections shown in Figure 1.

a. RQ1: Challenges faced by Indian farmers and opportunities for technological solutions

Over the past several decades, the share of agriculture and allied sectors, including forestry, logging, and fisheries, in the national GDP has steadily declined, decreasing from over 50% in the 1950s to approximately 14% in recent years [17]. Figure 3 illustrates that the sector’s projected growth rate remains significantly lower than the overall economic growth rate of 6%–7%. If current patterns continue, the agricultural sector’s contribution to GDP will keep on decreasing.

Figure 3:

The following sub-sections outline the major constraints currently limiting agricultural advancement in India.

b. RQ1: Technological advancements in Indian agriculture

As illustrated in Figure 4, the Indian agriculture industry is undergoing a paradigm change toward Agriculture 4.0, which is defined by the convergence of connectivity-driven systems, automation, and data analytics. Government-led programs like Digital India, Make in India, and Smart Villages, which together seek to digitize rural infrastructure and improve the effectiveness of agricultural production systems, encourage this transition. These initiatives have expedited the uptake of cloud computing, blockchain, IoT, and AI, creating a data-centric ecosystem for sustainable farming.

Figure 4:

Recent reports indicate that in pilot implementations across various agroclimatic zones, digital tools have increased average crop productivity by 10%–15%, showing quantifiable improvements in operational efficiency and revenue stability. Table 4 provides an overview of these applications. One prominent example is the AI-powered sowing application [29] created in collaboration with the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) and Microsoft. It uses cloud-based predictive analytics to identify the best sowing windows based on temperature profiles, rainfall patterns, and soil moisture. A 20% increase in production consistency and a 30% decrease in crop failure risk were found in field trials in Andhra Pradesh, demonstrating the importance of precise timing in agronomy. Research institutes and startups have become important forces behind the digital transition. Proximal Soilsens Technologies Pvt. Ltd. developed Soilsens [30], an invention funded by the Ministry of Electronics and Information Technology and the Department of Science and Technology. It reduces water waste by about 25% by using real-time soil parameter monitoring to deliver sensor-based irrigation advisories. In a similar vein, Jivabhumi [31] uses blockchain technology to guarantee transparent farm-to-market tracking for organic goods, reducing the abuse of middlemen and guaranteeing fair prices for farmers. A digital agritech platform called Gramophone [32] uses picture recognition algorithms to identify crop stress and suggest the best pest and nutrient management strategies. In the meantime, Gobasco [33] uses big data analytics and AI to optimize supply chains through post-harvest optimization, demand forecasting, and predictive logistics. CropIn [34], which offers farm-level forecasts, disease and pest alerts, and predictive yield analytics, is another example of AI-driven agronomic intelligence. Intello Labs [35] reduces subjective variability and increases agri-supply chain transparency by automating crop quality assessment at the point of aggregation using deep learning (DL) and computer vision. Agri-fintech [36] technologies have revolutionized agricultural financing by combining digital lending, crop insurance, and AI-enabled credit scoring to complement these production-side developments. By providing access to cash for technology adoption, risk mitigation, and long-term precision tool investment, these platforms increase financial inclusion for small and marginal farmers.

When taken as a whole, these innovations represent a major step forward for intelligent, resilient, and sustainable agriculture in India. A favorable environment for expanding AI-IoT technologies has been established by the collaboration of academic research, business innovation, and public policy, connecting traditional knowledge with digital intelligence. But for India’s agricultural digital revolution to be successful, systemic issues like digital literacy, affordability, and infrastructure shortages must be addressed.

c. RQ2: IoT’s journey in Indian farming: from remote villages to modern farms

Agriculture 4.0 has been made possible by IoT, which enables multi-layered connectivity between sensing devices, communication networks, and cognitive analytics systems. From a conceptual standpoint, it is a cyber-physical ecosystem made up of perception, network, and application layers that work together to monitor, control, and optimize agricultural processes in real time. It serves as a decision intelligence interface within this ecosystem, connecting digital analytics infrastructures with the physical farm environment to improve resilience, sustainability, and productivity throughout India’s agricultural value chain. Figure 5 displays IoT adaptation figures for India’s agricultural industry.

Figure 5:

Indian agriculture has shown quantifiable economic and environmental benefits using IoT-integrated DSS. In order to minimize Phytophthora infestans (late blight) outbreaks, IoT-based DSS that combines real-time weather forecasts with plant pathology models [37] has optimized fungicide scheduling for potato cultivation. These solutions improved the effectiveness of active substances and lessened their environmental impact while saving about US$500 per acre. Large-scale IoT deployments at the state level, such the Andhra Pradesh government’s IoT-enabled agricultural cycle management experiment [38], show scalability. The platform creates adaptive drought and irrigation warnings by combining GIS-based spatial data, soil health metrics, and multivariate climate inputs. The method improves spatiotemporal prediction accuracy and facilitates dynamic crop management by utilizing data assimilation frameworks that combine satellite imagery (NDVI, LST) with in-situ sensor measurements.

Recent advancements demonstrate how IoT and DL architectures are merging to enhance agrometeorological forecasting. To accurately anticipate rainfall and temperature anomalies, for example, gated recurrent units (GRU) have been utilized to process multi-sensor time-series data after missing-value regeneration, z-score normalization, and Fourier-based feature extraction [39]. In plant pathology, adaptive Gaussian filtering and contrast-limited histogram equalization have been used to improve the classification of disease symptoms from IoT-acquired photos using ResNet50 convolutional neural networks [40], which have been refined by transfer learning. The combination of data collection and inferential modeling for real-time agricultural intelligence is demonstrated by these combined IoT-DL frameworks. The integration of automation and renewable energy technology is further demonstrated by the installation of solar-powered IoT irrigation controllers [41]. In experimental projects throughout Maharashtra and Telangana, these systems reduce groundwater withdrawal and energy consumption by 20%–35% by using PID control algorithms and soil moisture-based actuation to deliver exact irrigation quantities. IoT integration and energy efficiency are positively correlated throughout the Yamuna Tributary Commercial Region, according to sustainability studies using directional distance functions and Malmquist-Luenberger productivity indices [42]. Distributed sensing of microclimatic variables with feedback loops for automated fertigation and irrigation is made possible at the operational scale by WSNs integrated with IoT controllers. In order to facilitate low-bandwidth communication, these networks use lightweight communication protocols like MQTT and CoAP and use star or mesh topologies, depending on the limitations of the terrain [43]. IoT-driven hydrological models use soil sensor data and FAO-56 Penman-Monteith formulae to predict crop water requirements (CWR) in arid and semi-arid locations, supporting sustainable irrigation management [44].

IoT is officially acknowledged as a driver of digital supply chain efficiency at the policy level in India’s “Doubling Farmers’ Income” objective [45]. IoT-based traceability solutions that combine blockchain technology, RFID tagging, and spectroscopic imaging enable real-time logistics monitoring and food quality authentication, lowering post-harvest losses, enhancing market transparency, and boosting customer confidence. The government’s focus on AI-IoT interoperability is encouraging the growth of smart agricultural ecosystems, in which mobile platforms function as digital twins of farms, connecting localized, context-aware advice with remote sensing data.

AI is also being used by India to promote agricultural modernization. Through the study of extensive agricultural information, AI-driven solutions provide data-driven suggestions and facilitate the early detection of biotic and abiotic challenges [46]. While AI improves telemetry image processing and precision agriculture by optimizing resource allocation and yield outcomes without replacing conventional practices, mobile-based expert systems help farmers with species recognition, soil diagnostics, and disease detection [47]. By enabling continuous data collection through IoT connectivity, AI further aids robotics and autonomous systems. Predictive decision-making is made possible by data mining, which transforms these datasets into useful insights [48]. To enhance food safety, traceability, and pathogen surveillance, cutting-edge technologies like blockchain and next-generation genomic sequencing are being investigated. Simultaneously, social media analytics are being employed more and more to promote agricultural innovation and comprehend consumer preferences [49].

Despite these developments, there are still several obstacles in the way of fully integrating AI and IoT in Indian agriculture. These consist of:

• Providing different farming communities with trustworthy, location-specific advising content.

• Making AI tools accessible in local languages and ensuring digital literacy in underprivileged areas.

• Using user input and longitudinal analysis to establish impact assessment frameworks for tracking adoption, behavioral change, and system efficacy [45,46,47,48,49].

d. RQ3: Tailoring AI algorithms for Indian crops and agricultural practices

The agricultural sector in India is remarkably diverse, with a wide range of crop systems, farming techniques, and agroclimatic zones. Table 5 shows the AI applications for Indian crops and agricultural practices. To improve production, sustainability, and resource efficiency while concurrently tackling pressing issues like water scarcity, managing pests and diseases, and climate-induced variability, the integration of AI has become increasingly important [50, 51]. However, because of contextual differences in farm size, infrastructure, and environmental variability, the majority of globally developed AI solutions are not immediately applicable to Indian settings. Therefore, Indian agriculture needs to develop localized and adaptive AI frameworks that prioritize crop-specific intelligence models, sustainable practices, and efficient resource management [52,53,54,55]. Representative AI algorithms designed for Indian crops and climate conditions are presented in Table 5. These include self-sufficient irrigation systems that are adjusted to local soil and weather conditions, improving the accuracy of water supply and irrigation scheduling predictions [56–57]. In a nation where more than 60% of agricultural land is still dependent on the monsoon, AI-based hydrological models enable precision water management by combining data from satellite imaging, soil moisture sensors, and weather stations [58–59].

Through customized, real-time advice, AI-driven DSS further empowers small and medium-sized farmers, who are the backbone of Indian agriculture. To maximize fertilizer application, crop selection, and pest control, these systems evaluate multi-modal data sources, such as soil health indicators, nutrient deficits, and monsoon forecasts [60,61,62,63,64,65,66,67]. Strong models for staple crops like rice, wheat, sugarcane, and pulses can be developed using region-specific datasets, guaranteeing agro-ecological compatibility and increased adoption rates [68,69,70]. Additionally, AI is essential for the early diagnosis of crop illnesses and pest infestations. DL models can precisely categorize disease symptoms based on texture, color, and morphological patterns when trained on photos taken by smartphones, drones, and IoT-enabled field cameras [72,73,74,75,76,77,78,79,80,81]. These models’ ability to learn adaptively is improved by the use of farmer-sourced feedback loops and institutional datasets. Additionally, AI-driven smart greenhouse systems optimize temperature, humidity, light intensity, and microclimatic parameters to maintain year-round precision farming and increase production [82,83,84].

Difficulties in applying AI to Indian agriculture are

• Data availability: Training scalable AI models for a range of farming situations is hampered by the lack of organized, high-quality, digital agricultural datasets [85].

• Integration of local expertise: AI systems must balance modern computational methods with indigenous and traditional farming expertise in order to attain practical relevance and adoption [86].

• Cost and accessibility: The high cost of AI-enabled technology prevents small and marginal farmers from adopting them, underscoring the need for affordable, mobile-friendly, and interoperable solutions that provide fair access to digital agriculture.

e. RQ4: Setting the stage for AI and IoT integration in Indian agriculture

The convergence of IoT sensing infrastructures and AI analytics—in which IoT devices gather real-time agricultural data and AI models process it to produce predictive insights and automated decision support—is becoming increasingly important to smart agriculture in India. In order to continuously monitor environmental and operational parameters including soil moisture, temperature, crop health, and irrigation levels, IoT technologies combine information and communication technologies (ICTs), unmanned aerial vehicles (UAVs), and agro-sensors. These sensors enable dependable connectivity across diverse agricultural landscapes in India by transmitting data via wireless communication networks such as 5G, LoRaWAN, NB-IoT, Sigfox, ZigBee, and Wi-Fi [87]. IoT sensors serve as the data acquisition layer in real-world implementations, and cloud or edge-based AI models serve as the analytics layer. Together, these components enable intelligent applications like precision fertilizer management, automated irrigation scheduling, disease diagnosis, and animal monitoring. The sector-wise deployment of IoT technology in Indian agriculture as of 2020 is shown in Figure 6.

Figure 6:

Numerous national projects show how IoT and AI technologies work together in actual agricultural ecosystems. To improve price discovery, lessen the role of middlemen, and increase transparency in agricultural commodity trade, the e-National Agriculture Market (eNAM) platform combines IoT-enabled market data gathering with AI-driven analytics [88]. Similar to this, IBM and NITI Aayog’s joint AI-based crop yield prediction system makes use of multi-source datasets collected via digital and remote sensing infrastructures, such as soil health databases, satellite imagery from the Indian Space Research Organization (ISRO), and weather observations from the India Meteorological Department (IMD). Machine learning algorithms are used to analyze these data streams and produce predictive warnings for farmers about the best crops to plant, when to irrigate, and how to reduce the danger of pest or disease outbreaks [89].

Programs for agricultural risk management also demonstrate the integration of AI with IoT-based remote sensing technologies. Agricultural and environmental data are gathered by satellite telemetry, field sensors, and GIS monitoring tools under the PMFBY. AI algorithms are then used to evaluate agricultural damage and expedite insurance claim payouts [90]. Similar to this, the Pradhan Mantri Kisan Samman Nidhi (PM-KISAN) program makes use of extensive digital datasets from government databases and farm records, where AI-based analytical models assess socioeconomic trends and facilitate focused policy interventions for smallholder farmers [91].

The integrated deployment of IoT and AI technologies is further encouraged by government-sponsored innovation hubs. Startups are encouraged to create intelligent agricultural solutions through initiatives like AGRI-UDAAN 3.0, which is funded by the Department of Science and Technology (DST) and uses IoT sensing platforms to feed real-time field data into AI-powered DSS for crop monitoring and resource optimization [92]. Additionally, regional projects show how this convergence is put into practice. For instance, the Karnataka government’s partnership with Microsoft offers farmers AI-enabled advising tools that use cloud-based analytics and IoT-generated field data to provide forecasts on weather and crop health [93]. Similarly, the Maha Agri-Tech project, which was created by the National Remote Sensing Centre (NRSC) and the Maharashtra Remote Sensing Application Centre (MRSAC), combines geospatial datasets, satellite imagery, and ground-based sensing data with AI-driven analytics to generate location-specific agronomic insights that support crop monitoring, land-use assessment, and adaptive farm management techniques [94].

These examples show how IoT-enabled sensing, communication networks, and AI-based analytics combine to create an integrated digital agriculture ecosystem that supports precision farming, risk mitigation, and sustainable agricultural development in India through real-time data acquisition and intelligent processing.

a. Temporal analysis of SLR on recent advancements in sustainable agriculture

The temporal analysis of the 96 evaluated papers provides insightful information about how research has changed between 2005 and 2026, highlighting both recent and historical advances in the field of multidisciplinary sustainable agriculture. The sharp rise in publications starting in 2020, which makes up the bulk of the evaluated literature, is a clear pattern in the data. With 28 publications, 2024 proved to be the most productive year. There are 11 publications in 2023 and 6 in 2022. Publications from early 2025 indicate that the field is still gaining momentum. This increase is indicative of an increasing national and international focus on digital transformation in agriculture, most likely due to:

• Advances in AI models tailored for precision farming.

• Proliferation of affordable IoT-based sensing systems.

• Increased governmental and institutional focus on smart agriculture initiatives in India.

From 2005 to 2015, only nine publications were recorded, suggesting the emergence of foundational technologies but limited application in agriculture. The field gradually grew between 2016 and 2019, with nine papers showing an increasing understanding of the possibilities of digital technologies in agriculture, even though they haven’t been used extensively yet.

Over 80% of the evaluated literature was published after 2020, according to Table 6’s year-by-year publication breakdown, highlighting the study area’s current significance and rapid evolution. It reflects-

• A greater understanding of farming that is climate resilient.

• Attempts to address food security through data-driven farming methods.

• The government’s promotion of “Digital India” initiatives, which promote digital adoption at the local level.

Nonetheless, the majority of recent research indicates that the discipline is still in its early stages of growth. Many solutions are conceptual or pilot-scale, necessitating

• Large-scale, real-world validation.

• Integration with regional farming methods and socioeconomic conditions.

• Policy frameworks to encourage smallholder farmers to embrace AI-IoT.

• The study’s research gaps and future directions.

• Absence of longitudinal studies: The long-term effects of AI-IoT systems in agriculture are rarely examined in publications.

• Limited regional focus: A large portion of the literature is either generic or concentrates on isolated case studies, despite the Indian context.

• Interdisciplinary integration: Future studies must integrate ecological, sociocultural, and economic factors with technological progress.

b. RQ5: performance analysis of AI and IoT techniques in Indian agriculture

A thorough analysis of recent research (Table 7) shows that although the integration of AI and IoT in Indian agriculture has the potential to be revolutionary, a number of significant constraints still exist across application domains. Models like GRU [39] and ResNet50 [40] show good predictive ability in weather forecasting and disease prediction. However, their deployment is hampered by issues including variable performance across heterogeneous field conditions, limited scalability, and sensor reliability. Furthermore, the absence of farmer-friendly interfaces hinders uptake and lowers accessibility.

The results of fixed-effects panel data regression models [42] evaluating low-carbon agriculture methods are promising, but they frequently ignore regional variety. IoT-enabled technologies, like wireless sensor networks [41] and precision irrigation frameworks [43, 51], are technically effective but need long-term validation in a variety of agroclimatic zones and seamless integration with AI analytics.

Although DL models like VGG [60], YOLOv5 [61], and DarkNet53 [62] have demonstrated proficiency in crop and disease identification, they struggle to identify small, dimly lit, or early stage symptoms in uncontrolled situations. Conventional models, such as BR-TomatoCNN [80] and MobileNetV1 [79], perform poorly in multi-label classification tasks and are not robust outside of controlled setups. Despite the potential for precision treatment offered by automated spraying systems [64, 65], their limited synergy with real-time disease detection algorithms limits their practical effectiveness.

Due to insufficient integration of varied data types (e.g., soil, weather, socioeconomic inputs), yield prediction models like Extra Trees Regressor [70] perform poorly in a variety of contexts. Similar to this, important real-world factors like policy changes and climate variability are frequently left out of price forecasting models like ARIMA-SVM hybrids [72], which reduces forecasting accuracy. In addition to having trouble with multi-label disease classification, advanced vision models like ResNet50 [76], InceptionV2 [79], and Faster R-CNN [81] also lack extensive symptom datasets. Although they provide environmental management solutions, cultivation control systems employed in mushroom farming [84] are still underappreciated in terms of scalability and predictive control integration.

Although they provide great precision for autonomous systems, emerging technologies like LiDAR [95], electro-optical cameras, and S-DGPS have not yet been successfully incorporated into conventional agricultural gear. Furthermore, as mentioned by Rokade and Singh [83], different technical needs for indoor greenhouses vs outside conditions necessitate focused research.

Bio-inspired algorithms for soil treatment, irrigation, and pesticide spraying have started to be deployed regionally [68], but they haven’t been evaluated in greenhouse settings or underrepresented agroclimatic zones. Additionally, ineffective information transfer and inadequate communication infrastructure restrict the practicality of algorithms. Research on customized communication technologies that are in line with algorithmic and environmental contexts is warranted.

The results of attempts to improve model accuracy with physiological and environmental inputs are inconsistent. To increase the resilience of the model, input parameter standardization is essential. Furthermore, intelligent integrated pest management (IPM) systems present a viable option and merit more research in light of growing chemical resistance [96]. Technological improvement by itself is insufficient to address global food security and climate change, even while the integration of AI with enhanced sensing and GIS technologies is becoming increasingly crucial. To overcome financial and infrastructure obstacles, policy-level initiatives like subsidies, innovation awards, and startup incubators are essential. Notably, smallholders have little use for current AI solutions, which are still biased toward large-scale farming. This gap can be closed and long-term food security enhanced by promoting youth involvement in technology-driven small-scale farming.

The absence of consistent evaluation criteria for evaluating the effectiveness of AI and IoT-based agricultural systems is another significant restriction found in all of the analyzed research. To assess their models, several studies use a variety of criteria, including accuracy, mean average precision (mAP), F1-score, R² mean absolute error (MAE), root mean square error (RMSE), water or energy savings, and ROC-based validation. Although these indicators show the efficacy of different treatments, direct comparison between studies is difficult due to the variation in evaluation methodologies. For instance, irrigation and resource optimization systems prioritize water-saving percentages or energy efficiency, while DL models for crop disease detection frequently report classification accuracy or F1-scores. Economic forecasting models, on the other hand, rely on statistical indicators like R² or prediction error metrics. The capacity to methodically evaluate the relative performance, scalability, and practical application of various technological solutions is hampered by the absence of standard evaluation frameworks. Future studies should thus concentrate on creating unified evaluation measures and standardized benchmarking procedures that consider technical accuracy, resource efficiency, scalability, economic viability, and farmer usability. The creation of solid, compatible smart agriculture solutions would be aided by the establishment of such uniform evaluation standards, which would allow for more trustworthy cross-study comparisons.

Scalability, long-term maintenance, and real-world sustainability are still major obstacles to the actual use of AI and IoT systems in agriculture, in addition to algorithmic performance. Although several of the examined systems show excellent accuracy in small pilot deployments or controlled experimental settings, their performance and dependability at large farm scales are still mainly unknown. Complexities such as varied soil conditions, fluctuating weather conditions, network connectivity constraints, and sensor deterioration over time are introduced by large-scale agricultural contexts. For instance, to maintain long-term effectiveness, IoT-based sensor devices need regular maintenance, constant calibration, and a dependable power source, which could raise farmers’ operating expenses. Similarly, the practical reliability of AI models trained on small datasets may be diminished by their inability to generalize across various agroclimatic zones and crop kinds. System sustainability, which includes price, ease of maintenance, and compatibility with current farming practices, is another crucial factor. Many of the solutions available now depend on sophisticated hardware or cloud infrastructure, which may not be financially feasible for small and marginal farmers, especially in developing nations like India. Thus, scalable architectures, low-cost sensor networks, edge-based processing, and reliable models that can adjust to changing field conditions should be given top priority in future research. To guarantee that AI–IoT agricultural technologies go beyond prototype-level implementations into sustainable real-world adoption, long-term field testing, cross-regional validation, and farmer-centric system design will be crucial.