Can you recognize it? How size and movement affect recognizability of equirectangular projections from immersive videos

Abstract

It is a proven fact that the size and animation of thumbnails representing a video’s content can influence the searching performance of people looking for it online. With immersive video being an up-and-coming new style of video, the aim of this experiment is to determine whether these results also hold for equirectangular projections, which are the thumbnail-equivalent for this new type of video format. For this purpose, we define performance as the accuracy and efficiency of searching for a thumbnail representing a known part of an immersive video. An experiment was run with 53 participants who were tasked to recognize an equirectangular thumbnail representing a part of a previously seen video excerpt in a grid of similar thumbnails. The thumbnails’ sizes and animation were adjusted for each task. Our analysis shows that both size and animation affect performance, in terms of efficiency. The results on accuracy are inconclusive. For example, still thumbnails improved the efficiency by 5% to 25% compared to animated ones and the larger thumbnail sizes provided significantly better efficiency compared to small thumbnails. These results are relevant and useful in the endeavor to build better search engine interfaces for immersive video data.