Exploring the application of knowledge transfer to sports video data
The application of Artificial Intelligence (AI) and Computer Vision (CV) in sports has generated significant interest in enhancing viewer experience through graphical overlays and predictive analytics, as well as providing valuable insights to coaches. However, more efficient methods are needed that...
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2025-02-01
|
| Series: | Frontiers in Sports and Active Living |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fspor.2024.1460429/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1825206759900315648 |
|---|---|
| author | Shahrokh Heidari Gibran Zazueta Riki Mitchell David Arturo Soriano Valdez Mitchell Rogers Mitchell Rogers Jiaxuan Wang Ruigeng Wang Marcel Noronha Alfonso Gastelum Strozzi Mengjie Zhang Patrice Jean Delmas Patrice Jean Delmas |
| author_facet | Shahrokh Heidari Gibran Zazueta Riki Mitchell David Arturo Soriano Valdez Mitchell Rogers Mitchell Rogers Jiaxuan Wang Ruigeng Wang Marcel Noronha Alfonso Gastelum Strozzi Mengjie Zhang Patrice Jean Delmas Patrice Jean Delmas |
| author_sort | Shahrokh Heidari |
| collection | DOAJ |
| description | The application of Artificial Intelligence (AI) and Computer Vision (CV) in sports has generated significant interest in enhancing viewer experience through graphical overlays and predictive analytics, as well as providing valuable insights to coaches. However, more efficient methods are needed that can be applied across different sports without incurring high data annotation or model training costs. A major limitation of training deep learning models on large datasets is the significant resource requirement for reproducing results. Transfer Learning and Zero-Shot Learning (ZSL) offer promising alternatives to this approach. For example, ZSL in player re-identification (a crucial step in more complex sports behavioral analysis) involves re-identifying players in sports videos without having seen examples of those players during the training phase. This study investigates the performance of various ZSL techniques in the context of Rugby League and Netball. We focus on ZSL and player re-identification models that use feature embeddings to measure similarity between players. To support our experiments, we created two comprehensive datasets of broadcast video clips: one with nearly 35,000 frames for Rugby League and another with close to 14,000 frames for Netball, each annotated with player IDs and actions. Our approach leverages pre-trained re-identification models to extract feature embeddings for ZSL evaluation under a challenging testing environmnet. Results demonstrate that models pre-trained on sports player re-identification data outperformed those pre-trained on general person re-identification datasets. Part-based models showed particular promise in handling the challenges of dynamic sports environments, while non-part-based models struggled due to background interference. |
| format | Article |
| id | doaj-art-ec99059af59a4c47b600026cc351097a |
| institution | Kabale University |
| issn | 2624-9367 |
| language | English |
| publishDate | 2025-02-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Sports and Active Living |
| spelling | doaj-art-ec99059af59a4c47b600026cc351097a2025-02-07T06:49:41ZengFrontiers Media S.A.Frontiers in Sports and Active Living2624-93672025-02-01610.3389/fspor.2024.14604291460429Exploring the application of knowledge transfer to sports video dataShahrokh Heidari0Gibran Zazueta1Riki Mitchell2David Arturo Soriano Valdez3Mitchell Rogers4Mitchell Rogers5Jiaxuan Wang6Ruigeng Wang7Marcel Noronha8Alfonso Gastelum Strozzi9Mengjie Zhang10Patrice Jean Delmas11Patrice Jean Delmas12IVSLab, The University of Auckland, Auckland, New ZealandUNAM, Monterrey, MexicoRiki Consulting, Auckland, New ZealandUNAM, Monterrey, MexicoIVSLab, The University of Auckland, Auckland, New ZealandNAO Institute, The University of Auckland, Auckland, New ZealandIVSLab, The University of Auckland, Auckland, New ZealandIVSLab, The University of Auckland, Auckland, New ZealandOne New Zealand Warriors, Auckland, New ZealandUNAM, Monterrey, MexicoCentre for Data Science and Artificial Intelligence, Victoria University of Wellington, Wellington, New ZealandIVSLab, The University of Auckland, Auckland, New ZealandNAO Institute, The University of Auckland, Auckland, New ZealandThe application of Artificial Intelligence (AI) and Computer Vision (CV) in sports has generated significant interest in enhancing viewer experience through graphical overlays and predictive analytics, as well as providing valuable insights to coaches. However, more efficient methods are needed that can be applied across different sports without incurring high data annotation or model training costs. A major limitation of training deep learning models on large datasets is the significant resource requirement for reproducing results. Transfer Learning and Zero-Shot Learning (ZSL) offer promising alternatives to this approach. For example, ZSL in player re-identification (a crucial step in more complex sports behavioral analysis) involves re-identifying players in sports videos without having seen examples of those players during the training phase. This study investigates the performance of various ZSL techniques in the context of Rugby League and Netball. We focus on ZSL and player re-identification models that use feature embeddings to measure similarity between players. To support our experiments, we created two comprehensive datasets of broadcast video clips: one with nearly 35,000 frames for Rugby League and another with close to 14,000 frames for Netball, each annotated with player IDs and actions. Our approach leverages pre-trained re-identification models to extract feature embeddings for ZSL evaluation under a challenging testing environmnet. Results demonstrate that models pre-trained on sports player re-identification data outperformed those pre-trained on general person re-identification datasets. Part-based models showed particular promise in handling the challenges of dynamic sports environments, while non-part-based models struggled due to background interference.https://www.frontiersin.org/articles/10.3389/fspor.2024.1460429/fullartificial intelligencecomputer visiontransfer learningzero-shot learningplayer re-identificationRugby League |
| spellingShingle | Shahrokh Heidari Gibran Zazueta Riki Mitchell David Arturo Soriano Valdez Mitchell Rogers Mitchell Rogers Jiaxuan Wang Ruigeng Wang Marcel Noronha Alfonso Gastelum Strozzi Mengjie Zhang Patrice Jean Delmas Patrice Jean Delmas Exploring the application of knowledge transfer to sports video data Frontiers in Sports and Active Living artificial intelligence computer vision transfer learning zero-shot learning player re-identification Rugby League |
| title | Exploring the application of knowledge transfer to sports video data |
| title_full | Exploring the application of knowledge transfer to sports video data |
| title_fullStr | Exploring the application of knowledge transfer to sports video data |
| title_full_unstemmed | Exploring the application of knowledge transfer to sports video data |
| title_short | Exploring the application of knowledge transfer to sports video data |
| title_sort | exploring the application of knowledge transfer to sports video data |
| topic | artificial intelligence computer vision transfer learning zero-shot learning player re-identification Rugby League |
| url | https://www.frontiersin.org/articles/10.3389/fspor.2024.1460429/full |
| work_keys_str_mv | AT shahrokhheidari exploringtheapplicationofknowledgetransfertosportsvideodata AT gibranzazueta exploringtheapplicationofknowledgetransfertosportsvideodata AT rikimitchell exploringtheapplicationofknowledgetransfertosportsvideodata AT davidarturosorianovaldez exploringtheapplicationofknowledgetransfertosportsvideodata AT mitchellrogers exploringtheapplicationofknowledgetransfertosportsvideodata AT mitchellrogers exploringtheapplicationofknowledgetransfertosportsvideodata AT jiaxuanwang exploringtheapplicationofknowledgetransfertosportsvideodata AT ruigengwang exploringtheapplicationofknowledgetransfertosportsvideodata AT marcelnoronha exploringtheapplicationofknowledgetransfertosportsvideodata AT alfonsogastelumstrozzi exploringtheapplicationofknowledgetransfertosportsvideodata AT mengjiezhang exploringtheapplicationofknowledgetransfertosportsvideodata AT patricejeandelmas exploringtheapplicationofknowledgetransfertosportsvideodata AT patricejeandelmas exploringtheapplicationofknowledgetransfertosportsvideodata |