sponsored by an individual anonymous donor and

JOIN US for a series of 7 free 45-minute seminars presented by student winners of Best Paper/Best Student Paper Awards within one of the conferences at the 2020 Electronic Imaging Symposium. Each seminar offers a 15-20 minute talk followed by group discussion.

Unless otherwise noted, seminars take place every two weeks until October 7 on Wednesday, and run from 10:00-10:45 EDT / 15:00-15:45 BST / 16:00-16:45 CET.

Titles below link to the EI2020 proceedings paper for the research on which the talk is based. These papers—like all EI proceedings—are open access.

Registration* for seminars open 1-2 weeks before each talk. Details on connecting are provided to those who register. Space is limited, but all seminars will be recorded for later viewing. Those who register will be notified when the recording becomes available.


Best Student Paper Intelligent Robotics and Industrial Applications using Computer Vision 2020 Conference

Saving lives with CNNs: Detecting hidden explosives based on suspicious environmental changes

based on the paper
Real-time Small-object Change Detection from Ground Vehicles Using a Siamese Convolutional Neural Network 
Sander Klomp, Eindhoven University of Technology (the Netherlands)

In modern warfare, hidden explosives are an ever present threat to military personnel. In fact, roadside Improvised Explosive Devices or “IEDs” are one of the leading causes of casualties among military personnel in conflict zones. In this seminar I explain how a vehicle-mounted stereo-camera system combined with AI algorithms can detect a hidden explosive before it’s too late.

Sander Klomp received his BSc and MSc from the Eindhoven University of Technology (TU/e) with the designation Cum Laude. He is now pursuing a PhD at TU/e in collaboration with ViNotion, with a focus on efficient deep learning algorithms.

>>Video Recording 


Best Paper 3D Measurement and Data Processing 2020 Conference

Learning a CNN on multiple sclerosis lesion segmentation with self-supervision
Alexandre Fenneteau, Siemens Healthcare (France)

We propose a novel multiple sclerosis lesion segmentation method for MR images, based on Convolutional Neural Networks (CNNs) with partial self-supervision and studied the pros and cons of using self-supervision for the current segmentation task.

Alexandre Fenneteau is a PhD student of Siemens Healthcare France working in the I3M laboratory on applying deep learning methods to medical images. The present work was performed during his master’s degree in 2019 in computational biology with the Quinze-Vingts Hospital, XLIM Laboratory, and I3M Laboratory.

>>Video Recording 


Best Paper Image Quality and System Performance (IQSP) XVII Conference   

Camera System Performance Derived from Natural Scenes  
Oliver van Zwanenberg, University of Westminster (UK)

Camera system performance is traditionally measured from calibrated test charts. The Modulation Transfer Function (MTF) is a well-established performance measurement that is used to evaluate the faithfulness of the reproduced image with respect to the chart contents. Our aim is to produce reliable MTF measure using natural scene images. The research can lead to camera evaluation in vital applications of computer vision technologies, such as CCTV and self-driving vehicles, as well as improve the performance of Image Quality models.

Oliver van Zwanenberg has an art and science background, with education in photography and imaging science, receiving his BSc Photography and Digital Imaging at the University of Westminster (2017) in London, moving on to pursue his PhD in the same year. Outside of his research studies, Oliver remains a keen photographer.

     IQSP Best Paper Award sponsored by NVIDIA

>>Video Recording 


Best Paper Imaging and Multimedia Analytics in a Web and Mobile World 2020 Conference

LambdaNet: A Fully Convolutional Architecture for Directional Change Detection  
Bryan Blakeslee, Rochester Institute of Technology (US)

Change detection in image pairs has traditionally been a binary process, reporting either “Change” or “No Change.” In this paper, we present LambdaNet, a novel deep architecture for performing pixel-level directional change detection based on a four-class classification scheme. LambdaNet successfully incorporates the notion of “directional change” and identifies differences between two images as “Additive Change” when a new object appears, “Subtractive Change” when an object is removed, “Exchange” when different objects are present in the same location, and “No Change".

Bryan Blakeslee is a recent graduate of Rochester Institute of Technology’s computer engineering program. His areas of interest are deep learning and embedded systems.

Imaging & Multimedia Analytics in a Web & Mobile World Best Paper Award sponsored by HP, Inc.

>>Video Recording 


**NOTE: This seminar will run from 8:00-8:45 EDT / 13:00-13:45 BST / 14:00-14:45 CET 

Arnaud Darmont Memorial Best Paper Award Image Sensors and Systems (ISS) 2020 Conference

An Over 120dB Dynamic Range Linear Response Single Exposure CMOS Image Sensor with Two-stage Lateral Overflow Integration Trench Capacitors
Yasuyuki Fujihara, School of Engineering Tohoku University (Japan)

A prototype linear response single exposure CMOS image sensor exhibiting over 120dB dynamic range with 11.4Me- full well capacity and maximum signal-to-noise ratio of 70dB will be presented. The developed CMOS image sensor is highly adaptive to many applications in automotive, medical, factory automation fields with strong contrast of light illumination.

Yasuyuki Fujihara is a doctoral course student at Tohoku University, Japan. He is engaged in researches on CMOS image sensors especially for spectral imaging and likes building IoT gadgets..

Arnaud Darmont Memorial Best Paper Award (ISS) sponsored by Dolby & sensors

>>Video Recording 


Best Student Paper 3D Measurement and Data Processing 2020 Conference

Variable Precision Depth Encoding for 3D Range Geometry Compression
Matthew G. Finley, University of Iowa (US)

Recent advances in computing and imaging technology have allowed for the high speed, high precision acquisition of 3D range data applicable to fields such as medicine, communications, physical security, and entertainment. However, the large file sizes associated with raw 3D data has shown to be prohibitive to target applications. This talk will explore the need for 3D data compression and will introduce a novel method for compressing data in such a way that points of interest are well-preserved.

Matthew Finley is a PhD student in the University of Iowa’s Holo Reality Lab studying applied image and signal processing.

>>Video Recording 


Best Student Paper Material Appearance (MAAP) 2020 Conference 

Caustics and translucency perception
Davit Gigilashvili, Norwegian University of Science and Technology (Norway)

Have you ever observed vivid patterns of light projected onto a table by a wine glass or a crystal figure? These patterns carry interesting information about the material they are cast by. In this talk, I discuss these patterns as a potential cue for helping us judge the translucence of materials.
Davit Gigilashvili is a third year PhD student at the Norwegian University of Science and Technology studying perception of material appearance. His research interests include gloss and translucency perception, color vision, computer graphics, and image quality. His pre-pandemic hobbies were hiking and traveling, while cooking and baking have become his major post-pandemic hobbies.

MAAP Best Student Paper Award sponsored by HP, Inc.


*By participating in the seminar, you consent to your voice/image possibly becoming part of the recording of the event. Recordings will be posted for others to view and may be used by IS&T for promotional or other distribution purposes.