Deep Learning (or Machine Learning with Neural Networks) is a very popular approach for designing data-driven artificial intelligence (AI). It is used in many of our applications for object detection, image enhancement, data synthesis etc.

Latest publications: IEEE ICASSP (2021), ACM IoT (2020), BMVC (2019), EUSIPCO (2019)

Pattern analysis is an important part for image processing. Some of our works have focused on developping robust techniques (e.g. M-estimation, Hough Transform, L2) for pattern registration (e.g. ellipse fitting) and reconstruction (e.g. 3D reconstruction from multiple views using RGB and RGB-D cameras).

Publications: Pattern Recognition (PR) (2018), PR (2016), PRL (2014), DSP (2013), PR (2013), PAMI (2009),

Colour transfer aims at changing the colour feel of an image (target image) using the colour palette of another image (palette image). Colour transfer can also be applied to videos, and for creating dynamic effects on static images. We have proposed several approaches to solve this problem based on Optimal Transport and information theory (L2 with GMM) as cost functions to minimize between colour distributions.

Publications: MMSP (2020), EUSIPCO (2019), IMVIP (2019), CVIU (2019), CVIU (2007), ICCV (2005)

Object detection in road scene sequential images has many applications from mapping to autonomous driving. My PhD thesis (2001) investigated linear machine learning techniques (PCA) and histogram backprojections for object detection and recognition. Our more recent approaches use Convolutional Neural Networks (CNNs).

Publications: IMVIP (2020), Remote Sensing (2018), ICIP 2018, CVPR 2000, ICIP 2001

Using computer vision and deep learning to automate registration of assets for telecoms operators, utility companies and civil infrastructure providers, AIMapIT replaces manual site surveys with fast, accurate detection and geotagging of assets from street level imagery with 2m accuracy.

Climate change is affecting people and their way of life including transportation of goods and people. Our work (2019) proposes to analyse satellite imagery in combination with street view imagery posted on social media during flooding events, for assessing the passability of roads. More recently (2020), we propose to infer 3d shape of roofs from aerial imagery (e.g. for measuring solar energy potentials).

Latest publications: Remote Sensing (2020)
Signal Processing: Image Communication (2019)

• Github: Remote sensing 2020

Information, references and some slides used for research and teaching computer vision and machine learning (e.g. MU module CS410) are shared online.

• CVML Webpage