Implementation of tensor flow-based deep learning in the learning application of around things in English

English is the most widely used international language in communication and is the second language after Indonesian. English will be taught more easily and effectively to children from an early age. But it needs learning media that suits the needs of children’s growth. Learning media should be fun and able to get the child moving to explore the surrounding environment. This research aims to create an English learning application by applying Tensor flow machine learning that is able to recognize objects around us and package them in a learning application in which there is a game that is able to attract the attention of children to learn English. This research was built using the RUP (Rational Unified Process) system development method which consists of 2 dimensions. The first dimension consists of dynamic aspects in development consisting of inception, elaboration, construction, and transition phases. The second dimension represents static aspects which are grouped into 4 important elements, namely who is doing, what, how, and when. The machine learning process consists of 3 stages, namely data input (object image capture), preprocessing, and the training process. The results of this study are in the form of a learning media application that can recognize surrounding objects by applying Tensor Flow-based deep learning. The application consists of 2 parts, namely vocabulary learning and a quiz for the introduction of surrounding objects. Parents have a role to be able to add names and objects to the quiz in the application. The results of the research can be used as a learning medium according to the needs of the child.