Super Long Range CNN For Video Enhancement in Handball Action Recognition

This paper explores the fusion of Longterm Recurrent Convolutional Neural Network (LRCNN) for action classification and Single Image Super-Resolution Convolutional Neural Network (SRCNN) for enhancing image quality in handball video sequences. Building upon previous work utilizing Mask R-CNN and an Optical Flow-Based Method for handball action detection, our study extends the framework by integrating LRCNN for precise action classification and SRCNN for improving spatial resolution and image quality as a hybrid model, Super Long Range Convolutional Neural Network (SLRCNN). LRCNN accurately classifies handball actions, while SRCNN enhances visual fidelity, leading to higher Peak Signal-to-Noise Ratio (PSNR) values. Through thorough experimentation and analysis, we assess the effectiveness of SLRCNN in capturing and classifying subtle handball actions, advancing sports analytics in computer vision with a holistic approach optimizing both action recognition accuracy and image quality metrics.