|Topic:||Detection of particular objects in images|
|Supervisor:||Georgios Tolias, Ph.D.|
|Announce as:||Diplomová práce, Semestrální projekt|
|Description:||Recent advances in deep learning for computer vision have significantly improved the performance of object detection. This is the case for learning detection of generic object classes, such as bicycle, horse, chair, monitor, etc., while less attention is given on particular objects, i.e. this bicycle, this chair. This is exactly the focus of this project, to learn how to detect particular objects given one or few examples. The goal will be pursued by training Convolutional Neural Networks. A side goal of the project is to perform the comparison between such CNN-based approach and classical approaches for particular object recognition and detection. This is an essential comparison that is missing in the literature.
|Bibliography:||Osokin et al, , ECCV2020, OS2D: One-Stage One-Shot Object Detection by
Matching Anchor Features
Dwibedi, Misra, Hebert, ICCV'17, Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detection.
Ammirato, Fu, Shvets, Kosecka, Berg, arxiv'18, Target Driven Instance Detection