In order to give new teams an easy entry, we provide results of content analysis to all teams. The V3C1 and V3C2 dataset already come with segmentation information and includes shot boundaries as well as keyframes. Moreover, we provide resulting data from different content analysis steps (e.g., color, faces, text, detected ImageNet classes, etc.). The analysis data of V3C1 is available here and described in this article, the one for V3C2 can be found here. Also, the ASR data has been released here (many thanks to Luca Rossetto et al.)! Moreover, the SIRET team shared their shot detection network too (many thanks to Jakub Lokoc and his team)!
Shot Boundary Detection
Existing Browsing Tool
If you want to join the VBS competition but do not have enough resources to build a new system from scratch, you can start with and extend a simple lightweight version of SOMHunter, the winning system at VBS 2020. The system is provided with all the necessary metadata for the V3C1 dataset. https://github.com/siret/somhunter
Providing a solid basis for research and development in the area of multimedia management retrieval, vitrivr is a modular open-source multimedia retrieval stack which has been participating to VBS for several years. It’s flexible architecture allows it to serve as a platform for the development of new retrieval approaches. The entire stack is available from https://vitrivr.org/
VBSl uses the V3C1+V3C2 dataset in collaboration with NIST, i.e. TRECVID (i.e. with the Ad-Hoc Video Search (AVS) Task) , which consists of 17235 video files, amounting for 2300h of video content (2508113 master shots) and 2.3 TB in size. In order to download the dataset (which is provided by NIST), please complete this data agreement form and send a scan to firstname.lastname@example.org with CC to email@example.com and firstname.lastname@example.org. You will be provided with a link for downloading the data.
Previous VBS Tasks
Tasks from VBS2018-VBS2022 can be downloaded from this GitHub repository.