Evaluating Algorithms for Temporal Queries in Ad-Hoc Video Retrieval (Bachelor Thesis, Ongoing)
With the tremendous increase of video recording devices and the resulting abundance of digital video, finding a particular video sequence in ever-growing collections is a major research challenge. The Video Browser Showdown is a yearly competition of research retrieval systems to test their ability to retrieve content based on either a textual description or seeing the target video sequence. vitrivr is an open-source system for indexing and retrieving multimedia data based on its content which has been a fixture in the Video Browser Showdown for the past years, winning in 2017 and 2019.
Recent evaluations have shown that enabling the user to express temporal context increases both selectivity of the query and the rank of the desired item. This problem can be viewed from two perspectives: i.) Query formulation, i.e., how does the user express temporal context and ii.) rank aggregation, i.e., how are the results of the individual queries merged. This thesis aims to make a contribution in both regards. It creates a ground-truth set of temporal queries and their correct results based on the tasks of VBS 2020 and the remote evaluation of SIRET and vitrivr. Different algorithms shall then be evaluated both on their runtime and retrieval performance (rank of the desired result in the result list when it appears there for the first time).
Start / End Dates
2021/03/01 - 2021/06/30