Automatic image captioning with a CNN and an LSTM

Automatic captioning of an image is a popular problem in AI that connects image processing and computer vision with NLP. In this recipe, you will learn how to use a pre-trained generative model (known as Show and Tell) based on a deep recurrent neural network architecture that can be used to generate captions (complete sentences in a natural language describing the contents of an image). The model was trained with the objective to maximize the likelihood of the input caption texts given the input training images. im2txt is a TensorFlow implementation of the Show and Tell model that can take images as input and generate human-like captions that describe the image. The model was tested on more than 300,000 images. The model is an end-to-end deep neural network consisting of a CNN (used to learn the implicit...