20.500.12556/RUL-155777 A new dataset and a distractor-aware architecture for transparent object tracking Performance of modern trackers degrades substantially on transparent objects compared to opaque objects. This is largely due to two distinct reasons. Transparent objects are unique in that their appearance is directly affected by the background. Furthermore, transparent object scenes often contain many visually similar objects (distractors), which often lead to tracking failure. However, development of modern tracking architectures requires large training sets, which do not exist in transparent object tracking. We present two contributions addressing the aforementioned issues. We propose the first transparent object tracking training dataset Trans2k that consists of over 2k sequences with 104,343 images overall, annotated by bounding boxes and segmentation masks. Standard trackers trained on this dataset consistently improve by up to 16%. Our second contribution is a new distractor-aware transparent object tracker (DiTra) that treats localization accuracy and target identification as separate tasks and implements them by a novel architecture. DiTra sets a new state-of-the-art in transparent object tracking and generalizes well to opaque objects. visual object tracking tracking transparent objects deep learning distractors vizualno sledenje objektov sledenje prosojnih objektov globoko učenje true false true Angleški jezik Slovenski jezik Članek v reviji 2024-04-17 13:21:42 2024-04-17 13:21:42 2024-07-25 03:52:43 0000-00-00 00:00:00 2024 0 0 Str. 2729–2742 iss. 8 Vol. 132 2024 0000-00-00 Zaloznikova Objavljeno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.93:004.8 0920-5691 10.1007/s11263-024-02010-0 186509827 s11263-024-02010-0.pdf s11263-024-02010-0.pdf 1 8BC175B3C2D1FC955633F460E012BC4A 7d01c77d21fbe4419467d511784f64b0b756991b371b604aa4c7445a0cfbaa1c 1cc85fbc-4990-11ef-8f74-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=187967 https://link.springer.com/article/10.1007/s11263-024-02010-0 1 acee6e85-fcac-11ee-85a5-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=182867 Fakulteta za računalništvo in informatiko 0 0 0