The Computer Vision Center at ECCV 2022
The 17th European Conference on Computer Vision (ECCV), took place on October 23-27 in Tel Aviv, Israel. The Computer Vision Center contributed with the following papers:
- OCR-IDL: OCR Annotations for Industry Document Library Dataset, by Ali Furkan Biten, Rubèn Tito, Lluis Gomez, Ernest Valveny & Dimosthenis Karatzas
- MUST-VQA: MUltilingual Scene-text VQA, by Emanuele Vivoli (CVC visiting student), Ali Furkan Biten, Andres Mafla, Dimosthenis Karatzas & Lluis Gomez
- Doc2Graph: a Task Agnostic Document Understanding Framework based on Graph Neural Networks, by Andrea Gemelli (CVC visiting student), Sanket Biswas, Enrico Civitelli, Josep Llados & Prof. Dr. Simone Marinai
Furthermore, CVC organised the 'Text in Everything' workshop in collaboration with AWS AI Labs (Amazon) & Cornell University. The workshop aims to bring together relevant speakers and researchers working on various aspects of text understanding in vision applications. It is a full-day event comprising invited talks, oral and poster presentations of submitted papers and a special session for the proposed challenge, the Out of Vocabulary Scene Text Understanding (OOV-ST) competition.
The ECCV 2022 Challenge on OOV-ST aims to evaluate the ability of text extraction models to deal with out-of-vocabulary words. The competition was organised by CVC in collaboration with AWS AI Labs & Google Research.