6.3.2020 Data Science Seminar: Learning to Segment in the Images

by Yona Binti-Abd-Gaus (Computer Science)

Friday 6 Mar 2020, 16:00 → 17:00 Europe/London

OC218 (Odgen Centre)

A lot of research is focused on object detection and it has achieved
significant advances with deep learning techniques in recent years. 
In terms of security, particular interest lies in the automatic
detection and classification of weapons within X-ray security imagery
in the airport, and drone based platform, with camera have been fast
deployed for surveillance purpose. However, these algorithms are not
usually optimal for dealing with images captured by drone-based
platforms and X-ray imagery, due to various challenges such as
limited availability of such images, rapid view point change,
different scales and density of object distribution and occlusion an
so on. In this presentation, we present initial visualisation results
for detection of objects from drone and X-ray space with Faster R-CNN 
model as baseline architecture. We hope that this preliminary results
can boost the research and development in visual analysis of X-ray
and drone platforms.