Hybrid CNN Bi-LSTM neural network for Hyperspectral image classification presents an innovative architecture meant for classifying the complex nature of hyperspectral images, which hold a non-linear relation between their spectral data and materials. This paper introduces a hybrid neural network model that combines 3-D CNN, 2-D CNN, and Bi-LSTM for effective learning of spatial and spectral features while reducing the number of parameters.

• Hybrid model consolidates 3-D CNN and 2-D CNN for spatial and spectral feature extraction.
• Bi-LSTM aids in learning inter-layer information effectively.
• Experimental results showcase improved performance on multiple datasets.
• Reduction in parameter count contributes to efficiency without compromising on accuracy.
• Achievements include accuracies of 99.83%, 99.98%, and 100% on select datasets with 30% less parameters.

This development marks a significant milestone in the field of remote sensing and environmental monitoring, optimizing resource use while boosting classification accuracy. It demonstrates how integrating different neural network structures can address the challenges of hyperspectral data analysis. Read more.