IMVIP 2018

Key dates

Paper submission is now closed.

  • Notification of Acceptance: 6th July 2018
  • Camera-ready papers: 17th July 2018

Early Registration deadline extended to 20th July 2018.

Conference Co-chairs

Welcome to the homepage of the Irish Machine Vision and Image Processing Conference, which will take place from August 29th - 31st 2018 in Ulster University, Belfast.

IMVIP brings together theoreticians and practitioners, industrialists and academics, from the numerous related disciplines involved in the processing and analysis of image-based information.

We invite submissions to IMVIP 2018 of papers presenting novel research contributions or applications related to any aspect of computer vision or image processing.

IMVIP is the annual conference of the Irish Pattern Recognition and Classification Society, a member body of the International Association for Pattern Recognition (IAPR). IMVIP 2018 is a single-track conference emphasising both theoretical research results and practical engineering experience.

The conference attracts both national and international participation, and provides an excellent forum for networking with academics, companies, researchers and students working in the area.

The conference programme will include oral and poster presentations, invited keynote speakers from world leading experts, industry demonstrations, and prizes for both Best Paper and Best Student Paper.

Full papers will be subject to a double-blind review process by the programme committee. Contributions are sought in all aspects of image processing, pattern analysis and machine vision, including but not restricted to the following topics:

  • Machine Learning and Deep Learning
  • Robot Vision Systems
  • Object and Event Recognition
  • Semantic Analytics
  • Scene Analysis and Visualisation
  • Medical and Bio-medical Imaging
  • Biometrics and Forensic Science
  • Face and Gesture Recognition
  • Social media applications with images and videos
  • Cognitive and Biologically-inspired Vision
  • Image Segmentation and Enhancement
  • Data Clustering and Texture Analysis
  • Applications, Architectures and Systems Integration
  • Image & Video Representation, Compression and Coding
  • Active Vision, Tracking and Motion Analysis
  • Image/Shape Representation and Recovery


Prof. Hiroshi Ishikawa

Prof. Hiroshi Ishikawa

Director of the Computer Vision and Analysis Lab
Department of Computing and Engineering
Waseda University

Structured Prediction by Fully Convolutional Deep Neural Networks

Best known for its breakthrough on the classification problem of general images, deep learning is now increasingly applied to other, more complex tasks.

Whereas the output of a classification algorithm is a simple label, these tasks require more structured output such as an image.

In this talk, I will present some recent results in our research in image transformation problems such as colorization of grey-scale images and image completion.

A fully convolutional network is suited to such tasks as it can handle input and output images of any size without relearning.

Prof. Jiri Matas

Prof. Jiri Matas

Centre for Machine Perception
Department of Cybernetics
Czech Technical University

Multi-Class Multi-Instance Model Fitting in Computer Vision

Many computer vision problems can be formulated as data fitting. In multi-class multi-instance fitting, the input data is interpreted as a mixture of noisy observations originating from multiple instances of multiple model types, e.g. as k lines and l circles in 2D edge maps, as k planes, l cylinders and m point clusters in 3D laser scans, as multiple homographies or fundamental matrices consistent with point correspondences in multiple views of a non-rigid scene.

After reviewing the evolution of data fitting methods including the Hough Transform, RANSAC and PEARL, I will present a novel method, called Multi-X, for general multi-class multi-instance model fitting.

The proposed approach combines global energy minimization using alpha-expansion and mode-seeking in the parameter domain.

Multi-X outperforms significantly the state-of-the-art on the standard dataset, runs in time approximately linear in the number of data points at around 0.1 second per 100 points, an order of magnitude faster than available implementations of commonly used methods.

I will also show how to plug efficiently and effectively the energy term into RANSAC. The resulting GC-RANSAC employs graph cut as local optimization to achieve state-of-the-art results.

Prof. dr. Zeno Geradts

Prof. dr. Zeno Geradts

Senior forensic scientist at the Netherlands Forensic Institute, and Professor of Forensic Data Science, University of Amsterdam.

Machine Vision and Image Processing in Forensic Science

The forensic science community makes since the last decades much more use of machine vision. The algorithms and computing power have been improved.

Image processing is routinely used in cases of CCTV, and intelligent search by machine vision is also progressing.

An overview is given of the current state of the art, concerning face comparison as well as other biometric features such as posture, gait and hand and feet comparison.

In robbery cases sometimes hands are a feature that can be used.

For child abuse cases also the feet are sometimes used for additional evidence, although research on algorithms for feet comparison is rare.

The challenge in forensic science is also the combination of features and reporting of the evidence as a likelihood ratio.

Since deep learning algorithms are more widely used, and calculation speed is improving, the use of camera identification based on Photo Non Uniformity (PRNU) is also progressing.

Some expectations in combined analysis are given with big data analysis systems as well as faster results in forensic science based on these methods, for instance in the European Union Horizon2020 project ASGARD.

Currently there are also many efforts in the anti-forensic field, such as morphing of faces and forging PRNU.

Detection of those forgeries is a research topic that is expected to attract much more attention in the future.


Registration for IMVIP 2018 is open.

Registration costs are listed below for both early and standard registrations, and for both students and non-students.

All registration types cover attendance at the conference, tea/coffee during breaks, lunch on Thursday 30th August, the conference welcome reception, and the conference dinner.

Early Registration (available up to 20th July 2018):

  • Early non-student: £240
  • Early student: £200

Standard Registration:

  • Non-student: £270
  • Student: £230

Book your place

Conference Programme

Wednesday 29th August

Wednesday Programme
12pm - 2pm Registration desk open
2pm - 2.15pm Welcome and Opening Remarks
2.20pm - 3.20pm

Keynote 1: Prof. Hiroshi Ishikawa

“Structured Prediction by Fully Convolutional Deep Neural Networks”

3.20pm - 3.50pm Tea Break  
Oral Paper Session 1
3.50pm - 4.10pm

1.1 Seam Carving for Content-aware Wide-angle Projection of Panoramic Photography

Darren Coughlan, Paul Cuffe

4.10pm - 4.30pm

1.2 Colour Correction for Stereoscopic Omnidirectional Images

Simone Croci, Mairéad Grogan, Sebastian Knorr, Aljosa Smolic - 4.50pm

1.3 Artifacts Reduction in JPEG-Compressed Images using CNNs

Fatma Albluwi, Vladimir A. Krylov, Rozenn Dahyot

4.50pm - 5.10pm

1.4 Visual and Semantic Feature Spaces for Zero-shot Image Decoding

Ben McCartney, Jesus Martinez-del-Rincon, Barry Devereux, Brian Murphy

5.10pm - 5.30pm

1.5 Security Analysis of the First Phase Mask in Double Random Phase Encryption

Lingfei Zhang, Thomas J. Naughton

MVIP 2018 Welcome Reception
5.30pm - 6.30pm

MVIP 2018 Welcome Reception

The Academy Restaurant, Ulster University Belfast campus

Thursday 30th August

Oral Paper Session 2

Oral Paper Session 2
9.10am - 9.30am

2.1 3D Point Cloud Segmentation using GIS

Chao-Jung Liu, Vladimir Krylov, Rozenn Dahyot

9.30am - 9.50am

2.2 LAP-based Tracking for Focal Adhesions

Katerina Lomanov, Jesús Martínez del Rincón, Paul Miller, Hugh Gribben

9.50am - 10.10am

2.3 Salient Obstacle Avoidance for Robotic Systems

Christopher Cooley, Sonya Coleman, Bryan Gardiner, Bryan Scotney

10.10am - 10.30am

2.4 Collaborative Dense SLAM

Louis Gallagher, John B. McDonald

Poster Pitches - Poster Session 1 - Coffee Break
10.30am - 11.20am

P1.1 Batch Normalization in the Final Layer of Generative Networks

Seán Mullery, Paul F. Whelan

P1.2 Template Matching for Head Pose Estimation

Shane Reid, Sonya Coleman, Dermot Kerr, Philip Vance, Siobhan O’Neill

P1.3 Impact Analysis and Tuning Strategies for Camera Image Signal Processing Parameters in Computer Vision

Lucie Yahiaoui, Jonathan Horgan, Senthil Yogamani, Ciaran Hughes, Brian Deegan

P1.4 Designing Objective Quality Metrics for Panoramic Videos based on Human Perception

Sandra Nabil, Raffaella Balzarini, Frederic Devernay, James Crowley

P1.5 The Effects of Noise on Deep Learning Architectures

James Hamm, Thomas J. Naughton

Oral Paper Session 3
11.20am - 11.40am

3.1 Neuromorphic Event-based Space-Time Template Action Recognition

Shane Harrigan, Dermot Kerr, Sonya Coleman, Pratheepan Yogarajah, Zheng Fang, Chengdong Wu

11.40am 12pm

3.2 Natural Gesture Extraction Based on Hand Trajectory

Naoto Ienaga, Bryan W. Scotney, Hideo Saito, Alice Cravotta, M. Grazia Busà

12pm - 12.20pm

3.3 Gesture Recognition with Thermopile Sensors

Matthew Burns, Philip Morrow, Chris Nugent, Sally McClean
12.20pm - 12.40pm

3.4 Towards Deep Learning-Based Emotion Detection for Affective Well-being

Rahul Sridhar, Haiying Wang, Huiru Zheng

Keynote 2
12.40pm - 2pm Lunch
2pm - 3pm

Keynote 2: Prof. Jiri Matas

“Multi-Class Multi-Instance Model Fitting in Computer Vision”

Oral Paper session 4
3pm - 3.20pm

4.1 Local Septenary Patterns for Breast Density Classification in Mammograms

Andrik Rampun, Bryan Scotney, Hui Wang, Philip Morrow

3.20pm - 3.40pm

4.2 Three-Dimensional Inverse-Weighted Interpolation of Cardiac Magnetic Resonance Images

James Fitzpatrick, Lorenzo Trojan, Niall Haslam, Kathleen Curran

3.40pm - 4.10pm Tea Break  
Oral Paper Session 5
4.10pm - 4.30pm

5.1 A Geometry-Sensitive Approach for Photographic Style Classification

Koustav Ghosal, Mukta Prasad, Aljosa Smolic

4.30pm - 4.50pm

5.2 Automatic Palette Extraction for Image Editing

Mairéad Grogan, Matis Hudon, Daniel McCormack, Aljosa Smolic

4.50pm - 5.10pm

5.3 Detection of Deepfake Video Manipulation

Marissa Koopman, Andrea Macarulla Rodriguez, Zeno Geradts

AGM & Dinner
5.30pm - 6.30pm IPRCS AGM

IMVIP 2018 Conference Dinner

Bert’s Jazz Bar, The Merchant Hotel, Belfast

Friday 31st August

Friday Programme
9.20am - 10.20am

Keynote 3: Prof. dr. Zeno Geradts

“Machine Vision and Image Processing in Forensic Science”

10.20am - 11.10am

Poster Pitches/Poster Session2/ Coffee Break

P2.1 Feature Selection, Reduction and Classifiers using Histogram of Oriented Gradients: How important is Feature selection?

Ryan Melaugh, Nazmul Siddique, Sonya Coleman, Pratheepan Yogarajah

P2.2 Evaluation of Residual Learning in Lightweight Deep Networks for Object Classification

Arindam Das, Senthil Yogamani

P2.3 Comparison of Texture Recognition Algorithms

Christina Sherly, Sonya Coleman, Dermot Kerr, Bryan Gardiner, Chengdong Wu

P2.4 Vehicle Blind Spot Monitoring using YOLO Object Detection

Maurice Barry, Seán Mullery

P2.5 Identification of Spatial Relationships in Arabic Handwritten Expressions using Multiple Fusion Strategies

Ibtissem Hadj Ali, Mohamed Ali Mahjoub

Oral Paper Session 6
11.10am - 11.30am

6.1 A Comparative Study of Face Re-identification Systems under Real-World Conditions

Glen Brown, Jesús Martínez del Rincón, Paul Miller

11.30am - 11.50am

6.2 Validation of Score-based Likelihood Ratio Estimation for Automated Face Recognition

Andrea Macarulla Rodriguez, Zeno Geradts, Marcel Worring

11.50am - 12.10pm

6.3 A Minkowski Distance-based Generalisation Method for Improving Centre Loss for Deep Face Recognition

Xin Wei, Hui Wang, Bryan Scotney, Huan Wan

12.10pm - 12.30pm

6.4 Face Morphing Detection

Ilias Batskos, Andrea Macarulla Rodriguez, Zeno Geradts

Best Paper Awards & Closing Remarks
12.30pm - 12.50pm Best Paper Awards & Closing Remarks

Conference Proceedings

All accepted papers will be published in the conference proceedings and will be available online under open access via the Ulster University Institutional Repository.

Student Bursaries

This year we will be awarding a number of student bursaries consisting of a conference registration fee waiver to support student participation in the conference.

These bursaries will be awarded on a competitive basis.

The eligibility criteria are that applicants must be a full-time or part-time student (confirmed by a letter from their supervisor), and must have either a long or short paper accepted at the conference.

To apply for a student bursary, download the application form and send your completed form by email to

Download application form

  • The closing date for applications will be 5pm GMT 20th July 2018.
  • Final decisions on award recipients will be communicated by 31st July 2018.

Please note that applicants are still required to register for the conference.

The mechanism for awarding the bursaries will be to refund the recipients their registration fee through the online registration system.

Venue & Travel

IMVIP 2018 will be held at the Belfast campus of Ulster University, situated in the artistic and cultural centre of the city, the Cathedral Quarter.

Although traditionally associated with art, the campus spans an increasing and exciting range of subjects including architecture, hospitality, event management, photography and digital animation.

Ulster University’s Belfast campus plays host to frequent fashion shows and art exhibitions and is an exciting place to study and work. It is also home to The Academy training restaurant.

Belfast Campus address

Ulster University
York Street
BT15 1ED
Northern Ireland

Car Parking

If travelling by car, there is a multi-storey car park – Q Park (St. Anne’s Square) – located just beside the campus.

Travelling to Belfast

Belfast has two airports, Belfast International Airport, and Belfast City Airport (also named the George Best Belfast City airport.) Both are served by many regional EU airports.

The travel time to Belfast by car or bus from the International airport is 30-40 minutes, and from the City George Best is around 10-15 minutes.

Dublin airport is a larger airport that serves many EU regional airports. A regular bus service links Dublin airport to Belfast, and takes approximately 1 hour 45 minutes.

Belfast International Airport

You can travel into Belfast either by taxi or bus; there is a regular service that takes you to the Europa/Great Victoria Street Bus Station. It is then a short walk to the Belfast campus (15 minutes).

Taxis can be pre-booked and there is also a taxi rank.

George Best City Airport

You can travel into Belfast either by taxi, bus or train; there is a regular service that takes you to the Europa/Great Victoria Street Bus Station. It is then a short walk to the Belfast campus (15 minutes).

Taxis can be pre-booked and there is also a taxi rank.

Dublin Airport

You can travel to Belfast from Dublin City Centre or Dublin airport by a direct bus that alights at Great Victoria Street/ Glengall Street in Belfast.

It is then a short walk to the Belfast campus (15 minutes).

There are two regular bus services from Dublin and Dublin airport to Belfast, and details can be found at the following:

You can also travel by train from Dublin City Centre (Connolly Street Station) to Belfast (Belfast Central Station).

The journey time is approximately 2 hours.

Travel by train directly from Dublin Airport to Belfast is not possible unless you first travel (by bus or taxi) to Dublin City Centre.

Staying in Belfast

Belfast is a vibrant, small city and you may wish to extend your stay for a day or two. Find out more at Visit Belfast.

The Ramada Encore and the Premier Inn hotels are both located within a 2-minute walk of the campus; the Premier Inn is slightly cheaper (around £60) than the Ramada Encore.

There are many other accommodation options available, including good quality B&B accommodation, and a range of 3* and 4* hotels. The Merchant Hotel (5*) is also located just a few minutes’ walk away.

3* Hotels

4* Hotels

5* Hotels

Instructions for Authors


Full papers

Full papers must be a maximum of 8 pages including references and appendices.

Full papers will be published in the conference proceedings, can be presented as a talk or a poster during the conference, and can be associated with demos also presented at the conference.

Short papers

Short papers must be a maximum of 4 pages including references and appendices.

Short papers are a convenient way of presenting in-progress research, student projects, or machine vision applications in industry.

Short papers will be published in the conference proceedings, can be presented as a poster during the conference, and can be associated with demos also presented at the conference.

Poster presenters will also have an opportunity to give a 2-minute overview of their work orally.

Template Files

All submissions must be in pdf format and must follow the formatting specified in the LaTeX template files found at the following link: The template provides instructions and examples for common document elements.

We also provide the following linked Microsoft Word template: imvip2018_Formatting_Instructions.docx

The Word template includes styles and layout settings that should assist in generating a pdf that is consistent with the LaTeX template output.

If using the Word template, please ensure that the final formatting abides by the LaTeX formatting as shown in the pdf file in


All papers must be submitted through the following link and should NOT include author names and affiliations:

Submission of a paper to the conference implies that, upon acceptance of the submission, at least one of the authors will register for the conference and present the paper.

Double-blind Review Process

All contributed papers will be subject to a double-blind review where both the authors and reviewers are anonymous.

Hence it is important that authors do not identify themselves in their own paper. In particular, submissions for review should not include author names or affiliations.