Caption generation is the task of describing an image with natural language. Previously, caption generation models worked on object detection models combined with templates that were used to generate text for detected objects. With all the advancements in deep learning, these models have been replaced with a combination of convolutional neural networks and recurrent neural networks.
An example is shown as follows:
Source: https://arxiv.org/pdf/1609.06647.pdf
There are several datasets that help us create image captioning models.
