Research on object segmentation and video object segmentation requires large amounts of annotated data, either to train models in the case of deep learning related solutions, or to verify the results of the developed methods. In this short paper we present an open source tool that aims to help in the task of annotating videos. This is both to improve the amount of annotated data and to ease the time for the annotation task. This tool is focused on underwater videos, incorporating specific image pre-processing algorithms and adding a model capable of suggesting fish detections. Moreover, the use of recent state of the art techniques permits the fast annotation of objects and the propagation of these annotations along the video sequence. We provide a public repository with the tool and through this paper we explain how we combined different techniques to improve the annotation times needed to create datasets.