AI-Powered Athlete Identification System for Real-Time Cricket Analytics
Uložené v:
| Názov: | AI-Powered Athlete Identification System for Real-Time Cricket Analytics |
|---|---|
| Autori: | B. Jyothi Chandra, R. R. Shantha Spandana, K. Bhaskar, K. Yatheendra |
| Zdroj: | International Journal of Applied Engineering and Management Letters. :132-140 |
| Informácie o vydavateľovi: | Srinivas University, 2025. |
| Rok vydania: | 2025 |
| Predmety: | cricket, LDA, Facerecognition, Deep learning, player detection, CNN |
| Popis: | In the realm of sports analytics, particularly in cricket, real-time and automated identification of players' faces is pivotal for performance tracking, strategy development, and audience engagement. However, recognizing players in dynamic match environments presents several challenges, including variations in facial pose, lighting conditions, background complexity, and occlusions. To address these issues, this project proposes an AI-powered athlete identification system that integrates advanced machine learning and deep learning techniques for accurate and robust face detection and recognition during live cricket matches. The system employs Linear Discriminant Analysis (LDA) for feature extraction, which enhances the discriminative power of facial features by reducing dimensionality and maximizing class separability. For effective face detection and classification, a Convolutional Neural Network (CNN) is utilized, leveraging pre-trained models to recognize players with high efficiency. To further improve recognition performance, the AdaBoost ensemble technique is incorporated. This method combines multiple weak classifiers to form a strong and accurate classifier, thereby increasing the robustness of player identification under varying conditions. Additionally, an enhanced VGG19-CNN model is implemented to optimize the face recognition process. This model architecture has been fine-tuned to adapt specifically to cricket match scenarios and has achieved an impressive accuracy of 95.5%. The fusion of LDA, CNN, AdaBoost, and the improved VGG19 model results in a powerful and resilient system capable of real-time athlete identification. The proposed system significantly contributes to automated cricket analytics by offering precise, fast, and scalable player recognition. It holds potential for wide applications in sports broadcasting, player performance monitoring, and data-driven decision-making. Even in complex and challenging match conditions, the system ensures reliable identification, paving the way for smarter and more engaging cricket analytics. |
| Druh dokumentu: | Article Journal |
| Jazyk: | English |
| ISSN: | 2581-7000 |
| DOI: | 10.47992/ijaeml.2581.7000.0238 |
| DOI: | 10.5281/zenodo.15803664 |
| DOI: | 10.5281/zenodo.15803665 |
| Rights: | CC BY URL: https://supublication.com/index.php/ijaeml/article/view/2079/1276 |
| Prístupové číslo: | edsair.doi.dedup.....8316e16ee9aeb23577a7b18b9dfe8248 |
| Databáza: | OpenAIRE |
Nájsť tento článok vo Web of Science | Abstrakt: | In the realm of sports analytics, particularly in cricket, real-time and automated identification of players' faces is pivotal for performance tracking, strategy development, and audience engagement. However, recognizing players in dynamic match environments presents several challenges, including variations in facial pose, lighting conditions, background complexity, and occlusions. To address these issues, this project proposes an AI-powered athlete identification system that integrates advanced machine learning and deep learning techniques for accurate and robust face detection and recognition during live cricket matches. The system employs Linear Discriminant Analysis (LDA) for feature extraction, which enhances the discriminative power of facial features by reducing dimensionality and maximizing class separability. For effective face detection and classification, a Convolutional Neural Network (CNN) is utilized, leveraging pre-trained models to recognize players with high efficiency. To further improve recognition performance, the AdaBoost ensemble technique is incorporated. This method combines multiple weak classifiers to form a strong and accurate classifier, thereby increasing the robustness of player identification under varying conditions. Additionally, an enhanced VGG19-CNN model is implemented to optimize the face recognition process. This model architecture has been fine-tuned to adapt specifically to cricket match scenarios and has achieved an impressive accuracy of 95.5%. The fusion of LDA, CNN, AdaBoost, and the improved VGG19 model results in a powerful and resilient system capable of real-time athlete identification. The proposed system significantly contributes to automated cricket analytics by offering precise, fast, and scalable player recognition. It holds potential for wide applications in sports broadcasting, player performance monitoring, and data-driven decision-making. Even in complex and challenging match conditions, the system ensures reliable identification, paving the way for smarter and more engaging cricket analytics. |
|---|---|
| ISSN: | 25817000 |
| DOI: | 10.47992/ijaeml.2581.7000.0238 |