Learning Observers’ Gaze Dynamics: An Efficient and Mobile Sport Scenery Recognition Pipeline

The project titled "Learning Observers’ Gaze Dynamics: An Efficient and Mobile Sport Scenery Recognition Pipeline" focuses on developing an automated system for recognizing sports-related images using machine learning techniques. The system utilizes a dataset of sport-related images from Kaggle, aiming to classify them into various sport categories. A key feature of the project is the integration of Grad-CAM (Gradient-weighted Class Activation Mapping), a technique that enhances the interpretability of the classification model by visualizing the important regions in the image that contribute to the model's decision. This allows users to better understand the rationale behind the predictions made by the model. The project aims to provide a solution that is optimized for mobile platforms, ensuring that the classification system is efficient and accessible, even on devices with limited resources. The system includes a user-friendly interface for registration, login, image upload, and prediction, ensuring ease of use for all users. By combining advanced image classification with Grad-CAM, this project not only improves the accuracy of sport image recognition but also addresses the need for model interpretability. The final solution provides an effective, transparent, and mobile-optimized approach to classifying sports-related images.

Keywords: Image Classification, Sport Scenery, Grad-CAM, Machine Learning, Flask, Dataset, Predictive System, Visualization, User Interface, Mobile-Friendly.