Year: 2026
Type: article
Abstract: To address the limitations of traditional cucumber disease recognition relying solely on image information with insufficient early warning accuracy, this paper proposes an intelligent early warning system for cucumber diseases based on multimodal data fusion. The system integrates leaf image data with environmental sensor data (temperature, humidity, light intensity, CO₂ concentration) and constructs an Attention Mechanism-based Convolutional Neural Network (AM-CNN). A multimodal dataset containing 12,000 samples across 7 disease categories was built by collecting cucumber leaf images and corresponding environmental parameters. Experimental results demonstrate that the AM-CNN model with multimodal fusion achieves an average recognition accuracy of 94.7%, which is 2.4 percentage points higher than the single-modal CNN model and 15.6-20.8 percentage points higher than traditional machine learning algorithms. The end-to-end response time is 1.6 seconds, meeting real-time monitoring requirements. This research provides a novel technical solution for precision disease warning in facility agriculture, with significant theoretical and practical value.
DOI: 10.63313/jcsft.9088
Authors: Fuxi Wei, Yuan Jiang, Hanzhang Luo, Guihong Wu
Institution: Baise University
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: hybrid
OpenAlex ID: W7166455809