Simplified Video Retrieval in Virtual Reality with vitrivr-VR
The use and application of virtual reality (VR) continue to grow as more advanced VR-capable hardware is developed. With VR hardware entering the mainstream, it becomes increasingly important to develop methods to support use cases previously performed using conventional user interfaces, such as desktop systems with keyboard and mouse. One very important such use case is video retrieval and browsing. Previous experiments have indicated that, while there appears to be a benefit to browsing videos in VR, the expression of textual queries remains a challenge. In this paper, we describe a new version of the vitrivr-VR virtual reality multimedia retrieval system, as it will participate in the video browser showdown (VBS) 2025. This new version of vitrivr-VR aims to overcome the two main challenges that have been identified in participation in previous installments of the VBS, namely the retrieval performance and the expression of textual queries, while continuing to leverage the advantages of video browsing in VR. We plan to overcome these challenges using a more performant retrieval backend (vitrivr-engine) and the use of newer VR-capable hardware (Apple Vision Pro) to enable easier text input.