We are often curious to find out what fruit is found in a tree when we are walking through parks or a botanical garden. Fruit classification is a complex activity that depends on the datasets used and the deep learning methods applied to them. This paper aims to present the steps by which we created a dataset with 262 fruits, and then how we created several models based on it. Evaluation and testing have shown us that the resource and models created can be useful to those who want to create applications that recognize fruits.