AI-Driven Tomato Crop Management System with Collaborative Learning
Harsha Vardhan N Dept. of EIE
RV College of Engineering Bengaluru, India harshavardhann.ei22@rvce.edu.in
Nitin Mammen Joy Dept. of EIE
RV College of Engineering Bengaluru, India nitinmammenjoy.ei22@rvce.edu.in
A S Nihal Dept. of IEM
RV College of Engineering Bengaluru, India asnihal.im22@rvce.edu.in
Sohan Raju M Dept. of ISE
RV College of Engineering Bengaluru, India sohanrajum.is22@rvce.edu.in
Aryann Guptha Dept. of CSE
RV College of Engineering Bengaluru, India aryannguptha.cs22@rvce.edu.in
Dr. S Venkatesh
Associate Professor, Dept. of EIE
RV College of Engineering Bengaluru, India venkateshs@rvce.edu.in
Abstract—Tomato cultivation in India faces persistent chal-lenges due to unpredictable weather patterns, soil variability, pest outbreaks, and inefficient post-harvest practices. To address these issues, this project proposes an integrated, AI-driven tomato crop management system that leverages energy-efficient IoT devices, computer vision, and federated learning to assist farmers with real-time, actionable insights. The system employs ESP32 microcontrollers connected to soil, pH, temperature, and light sensors, along with a vision-based assessment using CNN models deployed via Roboflow for disease detection and fruit ripeness evaluation. A user-friendly dashboard built with Vanilla JavaScript provides alerts, treatment suggestions, and crop health recommendations. Key innovations include solar-powered operation, low-cost implementation, and decentralized learning models that adapt to regional farming conditions while preserving data privacy. Field testing confirmed that the sys-tem accurately monitors environmental parameters, classifies diseases, and enhances decision-making for improved crop yield and resource efficiency. The results validate the practicality and scalability of the system for smallholder farmers across diverse agro-climatic zones.
Index Terms—Agriculture, IoT, DeepLearning, Sensors, Toma-toes, Automation, Vision, Disease, Ripeness, Dashboard
