Deep Learning for Cucumber Agriculture: A Hybrid CNN-SVM System for Disease Identification

This current study aims to develop a machine-learning model for the classification of cucumber diseases. This study categorises the six different leaf conditions. The dataset is supported by 6222 images which were taken from primary and secondary resources, taken from the cucumber field of the Punjab region and were pre-processed to remove the inconsistencies in the images such as backgrounds, lighting and unwanted objects. Features were extracted from the images using a CNN model in which the Support Vector Machines(SVMs) layer was integrated to enhance the capabilities by fine-tuning the model. The dataset was divided into training, validation and test-set for a comprehensive evaluation. The outcomes of this study highlight the model's effectiveness, with an average weighted F1-score of 77.16% and an overall accuracy of approximately 93.83%. The highest precision score for the model was 82.05 and the lowest was 73.93%. This research underscores the significance of early disease detection of cucumber leaf diseases. It also offers potential benefits for cucumber farmer in optimizing their crop health. These types of CNN models are helpful for farmers as they can save their time and cost. In future, we can develop more enhanced hybrid models which can give more accurate results.