Detail of the student project

Topic:Instance-level recognition with deep learning
Department:Katedra kybernetiky
Supervisor:Georgios Tolias, Ph.D.
Announce as:Diplomová práce, Bakalářská práce, Semestrální projekt
Description:The goal of this project is to design and train a recognition system for particular objects (recognize "this" chair, instead of "a" chair) with deep learning. Due to the large number of classes, the task is typically handled as an image retrieval problem. Each image is represented by a high dimensional descriptor and landmarks are recognized through the labels of the closest images in the descriptor space. The project will seek improvements to this pipeline by representing images with more than a single descriptor and by training pairwise verification models that can infer whether two images show the same object with higher certainty. Application: landmark recognition
Bibliography:Radenovic Tolias Chum, PAMI 2019, Fine-tuning CNN Image Retrieval with No Human Annotation
NohAraujoSimWeyandHan_ICCV2017_Large-Scale Image Retrieval with Attentive Deep Local Features
Responsible person: Petr Pošík