Wheat Leaf Disease Detection: A Hybrid CNN Based Approach With Multisource Data

Abstract

Healthy crop plays significant role for quality and good quantity productions of crops. Due to the population growth and limited farming land, it increases the demand of crop production. Wheat crop is one of the major crop of all over the world because it’s huge uses. Crop disease are one of the reasons for less production and bad quality of crop production. In early days when we not used machine learning and deep learning based methods to detect the crop disease, its time taken process, with the help of machine learning based methods we can detect the crop disease timely. Timely detection of crop disease help the farmers to remedy it on early stage. For timely and finest detection of disease in wheat leaf needed because lots of machine leaning based algorithm used for detection wheat leaf disease. So to take the advantages of various machine and deep learning techniques we can used the Hybrid based approach for detection the wheat leaf disease. Hybrid approach gives the flexibility to use more than one neural network architecture according to their strength and limitations. In this paper we propose CNN based hybrid approach that works on multisource data. Here we hybridize the CNN with some customization so that our model can give finest accuracy with multi source data. The multi-source data set will help to train our model that will give finest accuracy on light weight and real time data. The result suggest the most CNN performs better when we use original plus pre-process images and  with the help of data augmentation we can enhance the efficiency  from all models.