A novel deep neural network model using network deconvolution with attention based activation for crop disease classification

Accurate classification of crop diseases in its initial phase can help farmers to take necessary actions against the damage to their crops. The paper presents a Deep Learning (DL)-based crop disease classification approach that uses network deconvolution operation and attention-based activation function in each feature extraction layer. The existence of channel-wise and pixel-wise correlations in real-world images makes model training challenging. There is hardly any method available in the current literature that proposes the image correlation removal technique. The network deconvolution operation can efficiently remove both the correlations layer-wise. Hence, it is used in the proposed model. An attention-based activation function called AReLU is adopted in the model. The significance of AReLU activation function is to facilitate faster training. It can deal with gradient vanishing issue. The study considered Plant Village (PV), Tomato, and Grape datasets for performance evaluation. 80: 20 train-test split of the dataset was considered. The proposed model delivered significant results in comparison to other existing models, offering 100%, 99.27% and 99.10% classification accuracies and 99.88%, 99.06% and 99.01% F1-scores on Grape, PV and Tomato datasets respectively.

Automated summary

The paper introduces a Deep Learning-based approach for crop disease classification, using network deconvolution operation and attention-based activation function. The results show that the proposed model, utilizing network deconvolution operation and AReLU activation function, significantly outperforms other existing models in crop disease classification.