Android-Based Weed Identification and Herbicide Recommendation Using Convolutional Neural Networks

Ahmad Izzuddin (Universitas Panca Marga)
Ryan Prayuga Ardiansyah (Universitas Panca Marga)
Andrik Sunyoto (Universitas Panca Marga)
Dyah Ariyanti (Universitas Panca Marga)
Ira Aprilia (Universitas Panca Marga)

International Journal of Electrical and Intelligent Engineering
Vol 1, No 2 (2025)

Weed infestation reduces crop yield and quality, while inappropriate herbicide selection often limits effective control. This paper presents the design and implementation of an Android-based decision-support application for weed identification and herbicide recommendation using a smartphone camera. Weed images are classified using a lightweight Convolutional Neural Network with a MobileNetV2 architecture optimized for mobile deployment. Herbicide recommendations are generated using the Cosine Similarity method to associate identified weed characteristics with suitable control agents. The system is modeled using the Unified Modeling Language (UML) to ensure modularity and scalability. Experimental results show that the proposed CNN model achieves a classification accuracy of 96%. The integrated on-device image acquisition and intelligent recommendation enable practical field deployment, providing an efficient tool to support weed management decisions.