Related Experiment Video
Updated: Aug 16, 2026
06:41
Imaging and Analysis for Quantifying Maize (Zea mays) Abiotic Stress Phenotypes
Published on: March 28, 2025
Q-TriLSTM-Vision: a quantum-interference- augmented tri-stream LSTM for multi-label plant stress recognition on the
1School of Electronics Engineering, Vellore Institute of Technology, Vellore, Tamil Nadu, India.
Frontiers in Plant Science
|August 15, 2026
Summary
This study introduces Q-TriLSTM-Vision, a novel deep learning model for plant stress recognition in precision agriculture. It accurately detects plant diseases and deficiencies, improving crop productivity.
Area of Science:
- Agricultural Science
- Computer Science
- Artificial Intelligence
Background:
- Plant stress recognition is crucial for precision agriculture, yet current deep learning models face challenges with spatial dependencies and class imbalance.
- Existing Convolutional Neural Network (CNN), Transformer, and hybrid architectures struggle to differentiate visually similar stress symptoms.
Purpose of the Study:
- To propose Q-TriLSTM-Vision, a novel hybrid deep learning framework for enhanced plant stress recognition.
- To improve the accuracy and efficiency of detecting plant diseases, insect infestations, and nutrient deficiencies.
Main Methods:
- Developed Q-TriLSTM-Vision, integrating EfficientNet-B0, a tri-stream long short-term memory (TriLSTM) network, and a quantum-inspired interference gate.
- Utilized classical neural operations for quantum-inspired mechanisms, attention fusion, focal binary cross-entropy loss, weighted sampling, and per-class threshold calibration.
Main Results:
- Q-TriLSTM-Vision achieved high Macro-F1 scores (0.9127 on OLID-I, 0.9624 on PlantVillage, 0.8975 on PlantDoc).
- The model outperformed existing CNN, Transformer, and hybrid models, demonstrating lower Hamming loss and improved recall.
- Validation across multiple datasets and rigorous analysis confirmed the framework's robustness and effectiveness.
Conclusions:
- Q-TriLSTM-Vision offers an accurate, computationally efficient, and reliable solution for intelligent plant stress recognition.
- The proposed framework effectively addresses limitations in capturing spatial dependencies and handling class imbalance in multi-label classification.
- This advancement supports precision agriculture by enabling timely and precise identification of crop health issues.
