Crop Recommendation and Plant Disease Identification

Challenges appearing in agriculture are climate change, erosion, insufficient access to science and technology. All these resulted in plenty of unmanageable crops losses and diseases. Here we present a machine learning approach system that can suggest intelligent crops and accurately diagnose plant diseases. Thus, the system recommends crops based on the analysis of the data from a dataset of 2200 records where a number of the environmental parameters are used to do a systematic analysis of this data. We employed machine learning techniques of logistic regression, decision trees, Support Vector Machine, Multilayer Perceptron, Random forest to achieve such a goal. To identify the plant disease, we used convolutional neural network, in the form of VGG16 and EfficientNetV2S. To bring the peak of this performance, we trained these models by more than 70000 images of diseased plant leaves images in 38 categories, classified as per their overlapping nature. So, to increase the strength of prediction for all the models for the overall better performance of them, we further applied ensemble learning as stacking and weighted aver-aging techniques that is used to aggregate the power of prediction for a set of multiple models. These approaches reduce over fitting, enhance generalization to different datasets, and increase system accuracy and robustness. This can help to achieve a better solution that guides the process of decision making in agriculture. The work illustrates the potentials of advanced machine learning and ensemble algorithms to be utilized in crop recommendation and plant disease management optimization which serves as the tools to mitigate prime threats to agriculture.