{"id":{"repo_id":"qu-belfast","oai_identifier":"oai:pure.qub.ac.uk/portal:studenttheses/42d23663-5582-4a29-a390-a1b18961f7a5"},"canonical_url":"https://search.dev.ndltd.org/etd/qu-belfast/oai:pure.qub.ac.uk/portal:studenttheses/42d23663-5582-4a29-a390-a1b18961f7a5","repository":{"repo_id":"qu-belfast","name":"Queen's University Belfast","base_url":"https://pureadmin.qub.ac.uk/ws/oai"},"display":{"title":"Individual recognition from low quality and occluded images and videos using GAN","abstract":"Face recognition in the wild is one of the most interesting problems in the present world. Many algorithms with high performance have already been proposed and applied in real-world applications. However, the problem of recognising degraded faces from low-quality images and videos mostly remains unsolved. In recent times, we have seen a breakthrough in the perceptual quality of image enhancement with the recovery of the fine texture details and sharpness of degraded images. Several state-of-the-art algorithms, like ARCNN, IRCNN, SRGAN, etc., provide excellent results in the field of image enhancement. However, regarding face recognition, these algorithms fail miserably. Thus, we designed an algorithm which is specific to the enhancement of faces for better recognition performance. Thus, we present a versatile GAN capable of recovering facial features from degraded videos and images. The proposed method enhances the degraded faces by restoring their lost facial features rather than its perceptual quality, which ultimately leads to better performance of any existing face recognition algorithm. The limitation using a single image is that there is an upper bound to which we can enhance the images. This motivated us to develop the idea of synthesising the frontal face pose from multiple occluded images of the same identity, taken at different times and from different angles. This helps to further push the upper bound. Our network crowdsources useful information from all images and rejects information that is not useful for recognition. We build our generator on top of TPGAN and use the concept of a U-net discriminator which evaluates the output both at a global and local level, thus boosting the image to be consistent at a global and local level.","abstract_html":"Face recognition in the wild is one of the most interesting problems in the present world. Many algorithms with high performance have already been proposed and applied in real-world applications. However, the problem of recognising degraded faces from low-quality images and videos mostly remains unsolved. In recent times, we have seen a breakthrough in the perceptual quality of image enhancement with the recovery of the fine texture details and sharpness of degraded images. Several state-of-the-art algorithms, like ARCNN, IRCNN, SRGAN, etc., provide excellent results in the field of image enhancement. However, regarding face recognition, these algorithms fail miserably. Thus, we designed an algorithm which is specific to the enhancement of faces for better recognition performance. Thus, we present a versatile GAN capable of recovering facial features from degraded videos and images. The proposed method enhances the degraded faces by restoring their lost facial features rather than its perceptual quality, which ultimately leads to better performance of any existing face recognition algorithm. The limitation using a single image is that there is an upper bound to which we can enhance the images. This motivated us to develop the idea of synthesising the frontal face pose from multiple occluded images of the same identity, taken at different times and from different angles. This helps to further push the upper bound. Our network crowdsources useful information from all images and rejects information that is not useful for recognition. We build our generator on top of TPGAN and use the concept of a U-net discriminator which evaluates the output both at a global and local level, thus boosting the image to be consistent at a global and local level.","abstract_has_math":false,"creators":["Ghosh, Soumya Shubhra"],"institution":"Queen's University Belfast","degree_name":"Doctor of Philosophy","degree_level":"Doctoral Thesis","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Hua, Yang"],"committee_chairs":[],"committee_members":[],"year":2021,"date_issued":"2021-12","date_published":"2021-12","updated_at":"2026-07-24T03:56:04Z","subjects":["Face recognition","deep learning","neural network","generative adversarial networks","GAN","artifact removal","noise removal","super-resolution"],"languages":["eng"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["oai:pure.qub.ac.uk/portal:studenttheses/42d23663-5582-4a29-a390-a1b18961f7a5"],"render_values":[{"text":"oai:pure.qub.ac.uk/portal:studenttheses/42d23663-5582-4a29-a390-a1b18961f7a5","href":null,"code":true}]}]},"links":{"outbound_url":"https://pure.qub.ac.uk/en/studentTheses/42d23663-5582-4a29-a390-a1b18961f7a5","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Hua, Yang"]},{"key":"dc:contributor.sponsor","label":"Sponsor","values":["Anyvision Interactive Technologies Ltd","Anyvision (NI) Ltd"]},{"key":"dc:creator","label":"Author","values":["Ghosh, Soumya Shubhra"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2021-12"]},{"key":"dc:date.issued","label":"Date","values":["2021-12"]},{"key":"dc:publisher.department","label":"Dc Publisher Department","values":["School of Electronics, Electrical Engineering and Computer Science"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["Queen's University Belfast"]},{"key":"dc:relation.isreferencedby","label":"Dc Relation Isreferencedby","values":["https://pure.qub.ac.uk/en/studentTheses/42d23663-5582-4a29-a390-a1b18961f7a5"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["Doctoral Thesis"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["Doctor of Philosophy"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Face recognition","deep learning","neural network","generative adversarial networks","GAN","artifact removal","noise removal","super-resolution"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["oai:pure.qub.ac.uk/portal:studenttheses/42d23663-5582-4a29-a390-a1b18961f7a5","https://pure.qub.ac.uk/en/studentTheses/42d23663-5582-4a29-a390-a1b18961f7a5"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://pure.qub.ac.uk/files/250517923/PhD_Thesis_Final.pdf"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Face recognition in the wild is one of the most interesting problems in the present world. Many algorithms with high performance have already been proposed and applied in real-world applications. However, the problem of recognising degraded faces from low-quality images and videos mostly remains unsolved. In recent times, we have seen a breakthrough in the perceptual quality of image enhancement with the recovery of the fine texture details and sharpness of degraded images. Several state-of-the-art algorithms, like ARCNN, IRCNN, SRGAN, etc., provide excellent results in the field of image enhancement. However, regarding face recognition, these algorithms fail miserably. Thus, we designed an algorithm which is specific to the enhancement of faces for better recognition performance. Thus, we present a versatile GAN capable of recovering facial features from degraded videos and images. The proposed method enhances the degraded faces by restoring their lost facial features rather than its perceptual quality, which ultimately leads to better performance of any existing face recognition algorithm. The limitation using a single image is that there is an upper bound to which we can enhance the images. This motivated us to develop the idea of synthesising the frontal face pose from multiple occluded images of the same identity, taken at different times and from different angles. This helps to further push the upper bound. Our network crowdsources useful information from all images and rejects information that is not useful for recognition. We build our generator on top of TPGAN and use the concept of a U-net discriminator which evaluates the output both at a global and local level, thus boosting the image to be consistent at a global and local level."]},{"key":"dc:title","label":"Title","values":["Individual recognition from low quality and occluded images and videos using GAN"]}]}],"canonical_facts":{"dc:contributor.advisor":["Hua, Yang"],"dc:contributor.sponsor":["Anyvision Interactive Technologies Ltd","Anyvision (NI) Ltd"],"dc:creator":["Ghosh, Soumya Shubhra"],"dc:date":["2021-12"],"dc:date.issued":["2021-12"],"dc:description.abstract":["Face recognition in the wild is one of the most interesting problems in the present world. Many algorithms with high performance have already been proposed and applied in real-world applications. However, the problem of recognising degraded faces from low-quality images and videos mostly remains unsolved. In recent times, we have seen a breakthrough in the perceptual quality of image enhancement with the recovery of the fine texture details and sharpness of degraded images. Several state-of-the-art algorithms, like ARCNN, IRCNN, SRGAN, etc., provide excellent results in the field of image enhancement. However, regarding face recognition, these algorithms fail miserably. Thus, we designed an algorithm which is specific to the enhancement of faces for better recognition performance. Thus, we present a versatile GAN capable of recovering facial features from degraded videos and images. The proposed method enhances the degraded faces by restoring their lost facial features rather than its perceptual quality, which ultimately leads to better performance of any existing face recognition algorithm. The limitation using a single image is that there is an upper bound to which we can enhance the images. This motivated us to develop the idea of synthesising the frontal face pose from multiple occluded images of the same identity, taken at different times and from different angles. This helps to further push the upper bound. Our network crowdsources useful information from all images and rejects information that is not useful for recognition. We build our generator on top of TPGAN and use the concept of a U-net discriminator which evaluates the output both at a global and local level, thus boosting the image to be consistent at a global and local level."],"dc:identifier":["oai:pure.qub.ac.uk/portal:studenttheses/42d23663-5582-4a29-a390-a1b18961f7a5","https://pure.qub.ac.uk/en/studentTheses/42d23663-5582-4a29-a390-a1b18961f7a5"],"dc:identifier.uri":["https://pure.qub.ac.uk/files/250517923/PhD_Thesis_Final.pdf"],"dc:language":["eng"],"dc:publisher.department":["School of Electronics, Electrical Engineering and Computer Science"],"dc:publisher.institution":["Queen's University Belfast"],"dc:relation.isreferencedby":["https://pure.qub.ac.uk/en/studentTheses/42d23663-5582-4a29-a390-a1b18961f7a5"],"dc:subject":["Face recognition","deep learning","neural network","generative adversarial networks","GAN","artifact removal","noise removal","super-resolution"],"dc:title":["Individual recognition from low quality and occluded images and videos using GAN"],"dc:type":["Thesis"],"dc:type.qualificationlevel":["Doctoral Thesis"],"dc:type.qualificationname":["Doctor of Philosophy"]},"updated_at":"2026-07-24T03:56:04Z"}