Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Electrical and Electronic Engineering | en_US |
| dc.contributor.advisor | Siu, Wan-chi (EEE) | en_US |
| dc.contributor.advisor | Chan, Yuk-hee (EIE) | en_US |
| dc.contributor.advisor | Chan, Yui-lam (EEE) | en_US |
| dc.creator | Chan, Cheuk Yiu | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/14538 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Image enlightening with deep learning for playground lighting enhancement with action recognition | en_US |
| dcterms.abstract | This thesis addresses the challenges of automated action recognition in amateur soccer games, with a particular focus on overcoming the limitations posed by poor lighting conditions. While significant progress has been made in professional soccer analysis, amateur games present unique challenges including limited labeled data, diverse and unstructured video content, and frequently suboptimal lighting conditions. | en_US |
| dcterms.abstract | Our key contribution is a unique domain transfer approach that addresses these challenges through innovative low-light enhancement. The solution comprises two synergistic components: an action recognition model pre-trained on professional matches, and a specialized low-light image enhancement system that serves as a preprocessing step. To overcome the scarcity of training data for low-light conditions, we develop a novel diffusion-based framework for synthetic data generation, creating diverse and realistic low-light scenarios. | en_US |
| dcterms.abstract | Our primary contribution is a robust preprocessing framework that enhances visibility in poorly lit scenes while preserving crucial action details. This framework bridges the domain gap between amateur and professional footage, enabling more accurate action recognition across varying lighting conditions. | en_US |
| dcterms.abstract | Our research contributes new methodologies in low-light image processing and provides a theoretical framework for addressing the challenges of amateur sports analysis. The proposed approach lays groundwork for future developments in grassroots sports analytics, particularly in resource-constrained environments. | en_US |
| dcterms.extent | xiii, 119 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2026 | en_US |
| dcterms.educationalLevel | M.Phil. | en_US |
| dcterms.educationalLevel | All Master | en_US |
| dcterms.LCSH | Image processing -- Digital techniques | en_US |
| dcterms.LCSH | Human activity recognition -- Data processing | en_US |
| dcterms.LCSH | Deep learning (Machine learning) | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | open access | en_US |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/14538