Smart Harvest Diseases Identification using Image Processing

Abstract

Manual identification of leaf or fruit diseases is nearly impossible when it comes to large scale production, and it is a very time-consuming task that also requires a large amount of manpower. Hence, the existing technology is used where the disease identification is done within minutes. When a code is developed and deployed in any of the surveillance drones, a large-scale area can be covered in minutes and with less manpower. Here, in this proposal, there are two methods proposed. One is using the keywords and the other method is where a classifier is trained with the database available and the classifier with the highest accuracy is tested against the test set and a result is obtained. When there is a huge area of land, manually identifying the leaf diseases may require a lot of manpower and also be time-consuming. Hence, deploying this code in any sprayer drone will help in identifying the diseases within minutes. Along with disease identification, the particular longitude and latitude of the disease affected leaves’ location will be saved due to the fact that when the land is huge, the farmer need not go in search of the location; instead, a drone will fly to the exact location and the respective fertiliser will be sprayed. Before identification of leaf disease, a photo of the diseased leaves will be captured, due to which, by looking at the picture, the farmer can decide whether medication is necessary or not.