Modern agriculture has increasingly adopted Artificial Intelligence (AI), Internet of Things (IoT), machine learning, and cloud computing technologies to improve farming productivity and sustainability. Many previous studies highlight the use of IoT sensor networks to monitor soil moisture, pH levels, nutrient content, temperature, and humidity in real time. These systems help farmers optimize irrigation, fertilization, and field management practices. In parallel, computer vision techniques using Convolutional Neural Networks (CNNs) have been widely applied for automatic plant disease identification through leaf image analysis. Research has also introduced predictive analytics models for pest outbreaks, crop yield estimation, and climate-based decision support. Despite these advancements, many available solutions focus only on specific tasks and lack complete integration. Therefore, there is growing demand for smart agricultural platforms that combine multiple technologies into one efficient ecosystem.

Although several smart agriculture systems have been developed, many of them are limited in functionality, affordability, and accessibility. Existing solutions often concentrate only on disease detection, pest monitoring, irrigation control, or soil analysis individually rather than providing a complete farming support system. In addition, most advanced platforms require expensive hardware, strong internet connectivity, or technical expertise, making them unsuitable for small and medium-scale farmers. Another key limitation is the lack of localized AI models trained for regional crops, climate conditions, and farming practices. Few systems also provide multilingual interfaces, voice guidance, or explainable AI outputs that help farmers understand recommendations. Hence, there is a clear research gap for a cost-effective, user-friendly, localized, and integrated smart agriculture assistant.

Agriculture currently faces serious challenges due to climate change, unpredictable weather patterns, pest infestations, declining soil fertility, water scarcity, and increasing production costs. Traditional farming methods mainly rely on manual inspections and farmer experience, which can result in delayed disease identification, inefficient fertilizer use, poor pest control, and reduced crop yield. Farmers often lack access to timely information regarding soil health, plant growth conditions, and potential risks. This problem is more severe for small-scale farmers who cannot afford modern precision agriculture tools. Without intelligent support systems, farmers may experience crop losses, low profitability, and unsustainable farming practices. Therefore, there is a strong need for an affordable technology-driven solution that enables faster and smarter agricultural decision-making.

• To develop a localized AI system that identifies crop species and evaluates visible crop health conditions in Sri Lankan farming environments.
• To detect and classify plant diseases, pest infestations, and abnormal growth patterns for early intervention.
• To provide practical decision support for soil management, fertilizer selection, irrigation scheduling, and pest control.
• To deploy the complete solution as a lightweight web platform that remains accessible and useful for both rural and urban farming communities.

The main objective of this research is to design and develop an AI-powered Smart Agriculture Assistant that supports farmers in improving productivity, reducing crop losses, and promoting sustainable farming practices. Specific objectives include developing an IoT-based soil monitoring system to measure pH, moisture, and nutrient levels in real time. Another objective is to implement AI-based crop disease detection using image processing techniques. The research also aims to create predictive models for pest outbreaks and abnormal crop growth patterns. Additionally, the system will provide real-time alerts, treatment recommendations, and decision support through a mobile or web application. A further objective is to ensure the solution remains affordable, scalable, and practical for local farming communities.

The research follows a system development methodology that integrates hardware, software, and artificial intelligence components. Initially, soil sensors will be deployed in agricultural fields to collect parameters such as moisture, pH, and nutrient levels. Simultaneously, crop images will be captured using cameras or smartphones for disease and growth analysis. Collected data will be transmitted to a cloud platform where machine learning and deep learning models process the information. CNN models will classify plant diseases, while predictive algorithms analyze pest risks and crop growth abnormalities. The processed outputs will then be displayed through a user-friendly dashboard or mobile application with alerts, charts, and recommendations. Finally, field testing, farmer feedback, and performance evaluation will be conducted to validate system accuracy, usability, and real-world effectiveness.

System Overview

Workflow of Components