In recent years, rice production is mostly affected by rice plant leaf diseases due to the unawareness of suitable management strategies. The paddy leaves are regularly impacted by Brown spot and Bacterial blight diseases, which result in creating major loss to the farm owners. The naked-eye observation is used by the farmer to analyse the condition of paddy leaves, but, it takes more time and the accuracy of it is based on the observer. The naked-eye observation is generally difficult and it has a high possibility of human error. To overcome these drawbacks, a fast and suitable recognition system is required. Thus, appropriate methodologies are required for the determination of diseases in paddy leaf. The use of image processing is seen as a non-intrusive method that offers farmers a precise, economical, and trustworthy solution. Therefore, this research work, focused to provide the fast recognition system to detect leaf diseases in paddy crops.

There are several types of skin diseases, to protect and keep them healthy from these ailments; an effective and efficient diagnosis is required. One of the domains used by medical experts to diagnose severe class of skin disease is medical imaging. It is non-invasive way of diagnosis in which screen of the abnormal region performs first and then the dermatologist examines the subcutaneous structure and forecasts the severity of the lesion. One severe class of lesions is skin cancer, which is categorized as melanoma and non-melanoma. Most of the research has been performed on melanoma as yet and non-melanoma cancer diagnosis is still an untouched area. The cure rate of skin cancer is high, when diagnosed at an earlier stage. The proposed approach is applicable to gray scale or single channel images and the resultant output is binary images, and this can be compared easily with the available mask in the benchmark dataset. In addition to this, the APCNN proposal minimizes the requirement of post processing step for lesion boundary detection.