Abstract Crop detection using remote sensing imagery is a key component of precision agriculture, enabling large-scale monitoring of crop types with reduced manual intervention. However, conventional convolutional neural network (CNN)–based models often suffer from limited feature robustness and increased computational complexity when handling variations in scale, illumination, and noise. To address these challenges, this paper proposes a Scale Invariant Deep Convolutional Robust Feature Transformation–based Generalized Linear Regressive Classification (SIDCRFT-GLRC) model for accurate crop detection from remote sensing images. The proposed framework integrates scale-invariant deep feature learning with generalized linear regression to enhance discriminative feature extraction and classification efficiency. The model is evaluated using a Google Earth remote sensing image dataset, consisting of 50 to 500 multispectral images representing crops, trees, and plants. Experimental results demonstrate that the proposed SIDCRFT-GLRC model achieves a maximum classification accuracy of 97%, while reducing crop prediction time to 38 ms and achieving a false positive rate as low as 3% for 500 input images. Compared to existing ODE-RNN and WA-CNN approaches, the proposed method consistently outperforms them in terms of accuracy, computational efficiency, and misclassification reduction. These results confirm that SIDCRFT-GLRC provides an effective and scalable solution for crop detection using remote sensing images, supporting reliable agricultural monitoring with reduced computational complexity.
Devaki et al. (Mon,) studied this question.