Image understanding for road-traffic applications
by Siniša Šegvić | Nov 16, 2018
ORGANISATION
Name of the Organisation: University of Zagreb Faculty of Electrical Engineering and Computing
Organisation Short Name: UniZg-FER
Organisation Type: HES
Country: Croatia
ACTIVITY
Topic(s) of interest from 2018-2019 Transport Work Programme:- MG-2-7-2019: Safety in an evolving road mobility environment
- DT-ART-04-2019: Developing and testing shared, connected and cooperative automated vehicle fleets in urban areas for the mobility of all
- DT-ART-05-2020 - Efficient and safe connected and automated heavy-duty vehicles in real logistics operations
- DT-ART-06-2020: Large-scale, cross-border demonstration of connected and highly automated driving functions for passenger cars
Fields of Activity: Computer vision in images from the driver's perspective
Skills and Experise Offered: Robust semantic segmentation
- several submissions to Cityscapes and Pascal VOC 2012 challenges (Ladder-DenseNet)
http://openaccess.thecvf.com/content_ICCV_2017_workshops/w3/html/Kreso_Ladder-Style_DenseNets_for_ICCV_2017_paper.html
- runner-up at the Robust Vision Challenge at CVPR 2018 (ROB_LDN2)
http://www.robustvision.net/leaderboard.php?benchmark=semantic
- demo:
http://www.zemris.fer.hr/~ssegvic/pubs/kreso17semseg_stuttgart_00.mkv
- real-time performance (work in progress)
- dense out-of-distribution detection (work in progress)
Object detection and tracking
- demo:
http://www.zemris.fer.hr/~ssegvic/pubs/morsic18Linthescher.mp4
Keywords: computer vision, semantic segmentation, object detection and tracking
Previous FP Projects Participated: https://cordis.europa.eu/project/rcn/100294_en.html
CONTACT
Contact Person: Siniša Šegvić
Position in the Organisation: Full professor
Telephone: +38598313325
Email: sinisa.segvic@fer.hr