Evaluating segmentation in automatic captioning systems
We focused on evaluating the quality of the segmentation produced by the system, where decisions regarding the length, disposition, and display duration of the caption need to be taken so as to make the text easily readable.