Dear SIN participants, we have prepared additional annotations for iacc.1.A dataset, and we want to make them available to all participants of TRECVID SIN task. In total, we offer 178k positive and 588k negative annotations which are spread over 11k shots. You may download the annotations from http://dl.dropbox.com/u/45403487/BUT_annotations.tar.gz. The format is exactly the same as the one used for the collaborative annotations, so you should not have any problem using it. The annotations were gathered using a web-based tagging system which was developed at Brno University of Technology. The tagging system is partially described in the attached paper (will be presented at ACIVS 2012). Video shots were tagged one by one. Users marked multiple tags as positive or negative after watching a shot. The system allows typing the tags into a text box, selecting relevant tags from a list of automatically suggested tags, and copying positive tags from previous shots in a video. Videos were selected randomly for annotation and all shots in a video were usually annotated (except for very long videos). Total 300 man(student)-hours were spent. This approach to video-shot tagging provides more complete information about the shots and videos. We hope that the data will allow better correlation analysis of tags both within shots and within videos. The average number of positive and negative annotations for a shot is 17 and 54, respectively. You may use the annotations for any purpose without any restrictions. Please cite Hradis, M., Kolar, M., Kral, J., Lanik, A., Zemcik, P., Smrz, P.: Annotating images with suggestions - user study of a tagging system, In: ACIVS 2012 Proceedings, Brno, CZ, 2012, s. 1-12 http://www.fit.vutbr.cz/~ihradis/pubs.php.en?id=9990&shortname=1 Our servers are currently down due to maintenance. I hope they will be running soon. Let me know if you would like to get a login for the tagging system - you could view the annotations directly there. Please feel free to contact me if you have any questions or suggestions. Regards, Michal Hradią Department of Computer Graphics and Multimedia Faculty of Information Technology Brno University of Technology Boľetěchova 2, 612 66 Brno Czech Republic ihradis@fit.vutbr.cz http://www.fit.vutbr.cz/~ihradis/ +420 54114-1411