George Vogiatzis
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
George Vogiatzis has a total of 54 co-authors. Here are the top co-authors by number of shared publications:
- Roberto Cipolla
University of Cambridge
6 shared publications - Carlos Hernández
Google Inc
4 shared publications - Noa Garcia
Aston University UK
4 shared publications - Carlos Hernández
Toshiba Research Cambridge
3 shared publications Maria Chli
3 shared publications- Anders Bjorholm Dahl
Technical University of Denmark
2 shared publications
Latest Publications
- Open Access
- 24 September 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 330
A dual-aligned knowledge self-distillation framework for visible-infrared cross-modal person re-identification
- 24 September 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 330
Open AccessA dual-aligned knowledge self-distillation framework for visible-infrared cross-modal person re-identification
- 1 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 827-832
Multi-camera Torso Pose Estimation using Graph Neural Networks
Estimating the location and orientation of humans is an essential skill for service and assistive robots. To achieve a reliable estimation in a wide area such as an apartment, multiple RGBD cameras are frequently used. Firstly, these setups are relatively expensive. Secondly, they seldom perform an effective data fusion using the multiple camera sources at an early stage of the processing pipeline. Occlusions and partial views make this second point very relevant in these scenarios. The proposal presented in this paper makes use of graph neural networks to merge the information acquired from multiple camera sources, achieving a mean absolute error below 125mm for the location and 10° for the orientation using low-resolution RGB images. The experiments, conducted in an apartment with three cameras, benchmarked two different graph neural network implementations and a third architecture based on fully connected layers. The software used has been released as open-source in a public repository 1 .
- 1 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 827-832
Multi-camera Torso Pose Estimation using Graph Neural Networks
Estimating the location and orientation of humans is an essential skill for service and assistive robots. To achieve a reliable estimation in a wide area such as an apartment, multiple RGBD cameras are frequently used. Firstly, these setups are relatively expensive. Secondly, they seldom perform an effective data fusion using the multiple camera sources at an early stage of the processing pipeline. Occlusions and partial views make this second point very relevant in these scenarios. The proposal presented in this paper makes use of graph neural networks to merge the information acquired from multiple camera sources, achieving a mean absolute error below 125mm for the location and 10° for the orientation using low-resolution RGB images. The experiments, conducted in an apartment with three cameras, benchmarked two different graph neural network implementations and a third architecture based on fully connected layers. The software used has been released as open-source in a public repository 1 .
- Open Access
- 8 April 2020
- Published by Springer Nature
- Vol. 12035, 50-64
Variational Recurrent Sequence-to-Sequence Retrieval for Stepwise Illustration
We address and formalise the task of sequence-to-sequence (seq2seq) cross-modal retrieval. Given a sequence of text passages as query, the goal is to retrieve a sequence of images that best describes and aligns with the query. This new task extends the traditional cross-modal retrieval, where each image-text pair is treated independently ignoring broader context. We propose a novel variational recurrent seq2seq (VRSS) retrieval model for this seq2seq task. Unlike most cross-modal methods, we generate an image vector corresponding to the latent topic obtained from combining the text semantics and context. This synthetic image embedding point associated with every text embedding point can then be employed for either image generation or image retrieval as desired. We evaluate the model for the application of stepwise illustration of recipes, where a sequence of relevant images are retrieved to best match the steps described in the text. To this end, we build and release a new Stepwise Recipe dataset for research purposes, containing 10K recipes (sequences of image-text pairs) having a total of 67K image-text pairs. To our knowledge, it is the first publicly available dataset to offer rich semantic descriptions in a focused category such as food or recipes. Our model is shown to outperform several competitive and relevant baselines in the experiments. We also provide qualitative analysis of how semantically meaningful the results produced by our model are through human evaluation and comparison with relevant existing methods.
- 8 April 2020
- Published by Springer Nature
- Vol. 12035, 50-64
Open AccessVariational Recurrent Sequence-to-Sequence Retrieval for Stepwise Illustration
We address and formalise the task of sequence-to-sequence (seq2seq) cross-modal retrieval. Given a sequence of text passages as query, the goal is to retrieve a sequence of images that best describes and aligns with the query. This new task extends the traditional cross-modal retrieval, where each image-text pair is treated independently ignoring broader context. We propose a novel variational recurrent seq2seq (VRSS) retrieval model for this seq2seq task. Unlike most cross-modal methods, we generate an image vector corresponding to the latent topic obtained from combining the text semantics and context. This synthetic image embedding point associated with every text embedding point can then be employed for either image generation or image retrieval as desired. We evaluate the model for the application of stepwise illustration of recipes, where a sequence of relevant images are retrieved to best match the steps described in the text. To this end, we build and release a new Stepwise Recipe dataset for research purposes, containing 10K recipes (sequences of image-text pairs) having a total of 67K image-text pairs. To our knowledge, it is the first publicly available dataset to offer rich semantic descriptions in a focused category such as food or recipes. Our model is shown to outperform several competitive and relevant baselines in the experiments. We also provide qualitative analysis of how semantically meaningful the results produced by our model are through human evaluation and comparison with relevant existing methods.
- 24 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 22 (2) , 1264-1275
An Efficient Industrial System for Vehicle Tyre (Tire) Detection and Text Recognition Using Deep Learning
- 24 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 22 (2) , 1264-1275
An Efficient Industrial System for Vehicle Tyre (Tire) Detection and Text Recognition Using Deep Learning
- 23 November 2019
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 423-432
QuiltGAN: An Adversarially Trained, Procedural Algorithm for Texture Generation
- 23 November 2019
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 423-432
QuiltGAN: An Adversarially Trained, Procedural Algorithm for Texture Generation
- 1 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 4222-4229
A Deep Reinforcement Learning Agent for Traffic Intersection Control Optimization
- 1 October 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 4222-4229
A Deep Reinforcement Learning Agent for Traffic Intersection Control Optimization
- 1 August 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1074-1079
Vehicle tire (tyre) detection and text recognition using deep learning
This paper presents an industrial system to read text on tire sidewalls. Images of vehicle tires in motion are acquired using roadside cameras. Firstly, the tire circularity is detected using Circular Hough Transform (CHT) with dynamic radius detection. The tire is then unwarped into a rectangular patch and a cascade of convolutional neural network (CNN) classifiers is applied for text recognition. We introduce a novel proposal generator for localizing the tire code by combining Histogram of Oriented Gradients (HOG) with a CNN. The proposals are then filtered using a deep network. After the code is localized, character detection and recognition are carried out using two separate deep CNNs. The system presents impressive accuracy and efficiency proving its suitability for the intended industrial application.
- 1 August 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1074-1079
Vehicle tire (tyre) detection and text recognition using deep learning
This paper presents an industrial system to read text on tire sidewalls. Images of vehicle tires in motion are acquired using roadside cameras. Firstly, the tire circularity is detected using Circular Hough Transform (CHT) with dynamic radius detection. The tire is then unwarped into a rectangular patch and a cascade of convolutional neural network (CNN) classifiers is applied for text recognition. We introduce a novel proposal generator for localizing the tire code by combining Histogram of Oriented Gradients (HOG) with a CNN. The proposals are then filtered using a deep network. After the code is localized, character detection and recognition are carried out using two separate deep CNNs. The system presents impressive accuracy and efficiency proving its suitability for the intended industrial application.
- 8 June 2019
- Published by Springer Nature
- p. 749-755
Traffic3D: A Rich 3D-Traffic Environment to Train Intelligent Agents
- 8 June 2019
- Published by Springer Nature
- p. 749-755
Traffic3D: A Rich 3D-Traffic Environment to Train Intelligent Agents
- Open Access
Learning Non-Metric Visual Similarity for Image Retrieval
Measuring visual similarity between two or more instances within a data distribution is a fundamental task in image retrieval. Theoretically, non-metric distances are able to generate a more complex and accurate similarity model than metric distances, provided that the non-linear data distribution is precisely captured by the system. In this work, we explore neural networks models for learning a non-metric similarity function for instance search. We argue that non-metric similarity functions based on neural networks can build a better model of human visual perception than standard metric distances. As our proposed similarity function is differentiable, we explore a real end-to-end trainable approach for image retrieval, i.e. we learn the weights from the input image pixels to the final similarity score. Experimental evaluation shows that non-metric similarity networks are able to learn visual similarities between images and improve performance on top of state-of-the-art image representations, boosting results in standard image retrieval datasets with respect standard metric distances.
Open AccessLearning Non-Metric Visual Similarity for Image Retrieval
Measuring visual similarity between two or more instances within a data distribution is a fundamental task in image retrieval. Theoretically, non-metric distances are able to generate a more complex and accurate similarity model than metric distances, provided that the non-linear data distribution is precisely captured by the system. In this work, we explore neural networks models for learning a non-metric similarity function for instance search. We argue that non-metric similarity functions based on neural networks can build a better model of human visual perception than standard metric distances. As our proposed similarity function is differentiable, we explore a real end-to-end trainable approach for image retrieval, i.e. we learn the weights from the input image pixels to the final similarity score. Experimental evaluation shows that non-metric similarity networks are able to learn visual similarities between images and improve performance on top of state-of-the-art image representations, boosting results in standard image retrieval datasets with respect standard metric distances.
- 1 February 2019
- Published by Elsevier in Image and Vision Computing
- Vol. 82, 18-25
Learning non-metric visual similarity for image retrieval
- 1 February 2019
- Published by Elsevier in Image and Vision Computing
- Vol. 82, 18-25
Learning non-metric visual similarity for image retrieval