Related Experiment Video
Updated: Jul 2, 2025
06:41
Imaging and Analysis for Quantifying Maize (Zea mays) Abiotic Stress Phenotypes
Published on: March 28, 2025
800
Using wavelet transform and hybrid CNN - LSTM models on VOC & ultrasound IoT sensor data for non-visual maize disease
Theofrida Julius Maginga1, Emmanuel Masabo1, Pierre Bakunzibake1
1African Centre of Excellence in Internet of Things (ACEIoT) - University of Rwanda (UR), Rwanda.
Heliyon
|February 29, 2024
Summary
Early detection of Northern Leaf Blight (NLB) in maize is now possible using Internet of Things (IoT) sensors, identifying disease 4-5 days before visual symptoms. This innovation aids food security by protecting crop yields.
Area of Science:
- Agricultural Science
- Plant Pathology
- Sensor Technology
Background:
- Northern Leaf Blight (NLB) significantly reduces maize yield, with traditional detection taking 14-21 days.
- Early disease identification is critical for food security, especially in regions like Sub-Saharan Africa.
Purpose of the Study:
- To develop a novel, rapid detection method for NLB in maize.
- To utilize Internet of Things (IoT) sensors for pre-symptomatic disease detection.
Main Methods:
- Employed Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) models.
- Analyzed non-visual data: Total Volatile Organic Compounds (VOCs) and ultrasound emissions.
- Developed a hybrid CNN-LSTM model for VOCs and an LSTM model for ultrasound anomalies.
Main Results:
- The hybrid CNN-LSTM model achieved an F1 score of 0.96 and an AUC of 1.00 for VOC classification.
- The LSTM model demonstrated 99.98% accuracy in detecting ultrasound anomalies.
- Disease detection was achieved as early as 4-5 days post-infection.
Conclusions:
- IoT sensors offer a promising approach for early plant disease detection before visual symptoms manifest.
- This technology can lead to advanced disease prevention strategies, enhancing agricultural resilience.
- Future research will focus on IoT deployment optimization and field validation.
