Survey Paper On AI Based Sports Highlight Generation For Social Media

The rapid growth of social media platforms has transformed how sports content is consumed, with short, engaging highlight clips becoming a cornerstone of fan interaction. Traditional manual editing of highlights is time-consuming and inefficient, prompting the adoption of artificial intelligence (AI) to automate this process. This survey paper explores the state-of-the-art in AI-based sports highlight generation tailored for social media, focusing on techniques, applications, and performance evaluation. We review key advancements in video analysis (e.g., object detection, event segmentation), audio processing (e.g., crowd noise analysis), and multimodal approaches that integrate visual, auditory, and textual data to identify impactful moments. The paper examines prominent systems like WSC Sports and SPNet, highlighting their contributions to real-time and post-processed highlight creation. To assess these technologies, we conducted experiments using Google Colab, testing models such as 3D-CNN and pretrained Video-LLaMA on a sample dataset (SoccerNet). Results, visualized through tables and graphs, reveal high precision (up to 87%) and recall (up to 82%) in detecting key events, though challenges like real-time processing and subtle moment recognition persist. Social media demands—short duration, high engagement, and personalization—are analyzed, alongside practical applications such as fan engagement and monetization. However, technical limitations (e.g., dataset bias) and ethical concerns (e.g., privacy in crowd footage) remain hurdles. This survey underscores AI’s transformative potential in sports media, offering insights into current capabilities and future directions, including generative AI and immersive technologies like AR/VR. By synthesizing literature, experimental findings, and practical implications, this paper provides a roadmap for researchers and practitioners aiming to enhance sports highlight generation for the dynamic landscape of social media.