Progress on Deep Learning Models for Plant Disease Detection: A Survey

Deep Learning (DL) and Artificial Intelligence (AI) are well matured sufficiently and in use in multiple domains of applied sciences and engineering to great success. Research interest in DL has increased over the years and this is because of the great success brought about by it. Another reason for the increase in interest has been the enormous amount of data available and the computing power at our disposal to harness these datasets. As a direct result of the current research interest, different DL models and tools are being created every day to maximise the efficiency of use across multiple domains. This paper does a systematic review of the use of DL models for detecting plant diseases. Images of diseased plant leaves are processed using these DL models to achieve timely and accurate diagnosis in a way that can scale without human effort. The survey shows that various DL networks have been used to detect plant diseases. DL has also been used to check the severity of diseases in different plant species. Plant detection could be an error-prone, time-consuming process and destructive potentially if done in error or too late. This paper shows the current state of DL in plant disease diagnosis and its limitations and challenges that can encourage researchers to investigate further improvements.