Generating accurate captions for an image has remained one of the major challenges in Artificial Intelligence with plenty of applications ranging from robotic vision to helping the visually impaired.Long term applications also involve providing accurate captions for videos in scenarios such as security system.The aim is to build an optimal system which can generate semantically and grammatically accurate captions for an image and also to suggest captions which can be used on social media platforms.In this system, images are preprocessed and captions are generated. Then the image features are extracted using Resnet50.Then the captions are generated word by word using LSTM. The application hopes to be useful for visually impaired people and to be useful for generating the social media captions which can be used on various social media platforms.