Dragon fruit is also known as red pitaya, a fruit of the cactus genus. Dragon fruit is known for its nutritional value and its high antioxidant capability. Dragon fruits can be cultivated all around the year helping the country's economy. But some diseases like white spot, red spot etc. can be destroy the crops entirely as it is new in our country farmer are not aware these diseases yet. There is not much noticeable research going on regarding this fruit, people are focusing on the seasonal fruits like Mango, Jack fruit etc. In this work we presented a model which can classify diseases from dragon fruit leaves. There is a lack of resources for dragon fruit disease detection. Therefore, as a part of this work we have collected 1732 images of dragon fruit leaves from different fields which then divided in train, test and validation and used in this work. We have used a total of 5 models in this study which are CNN, CNN+SVM, CNN+Fusion, CNN+Attention, CNN+Transfer Learning. Among these models CNN+Transfer Learning achieved the highest accuracy of 92.83%. This paper is the first publicly available paper on dragon fruit leaf disease detection using hybrid transfer learning model to the best of our knowledge.