Computer Vision: page flipping detection

Through widespread application of cameras, modern CPUs and GPUs computer vision has become an amazingly popular tool to monitor, automate processes in different fields from medicine to agriculture. One of the niche applications of the machine learning on the images is to detect whether a page is flipped or not. Page flipping detection can be utilized for user experience optimization, content analytics, advertisements and monetization, user research and testing, as well as fraud detection in various contexts and industries.

2D Convolutional Neural Network

A 2D CNN was used for page flipping detection due to its ability to analyze two -dimensional input data, such as images or video frames. In the context of page flipping detection, the 2D CNN can process consecutive frames or images captured during the flipping process to identify the occurrence and characteristics of page flips. By learning spatial patterns and temporal dependencies from these sequential frames, the 2D CNN can effectively detect and track page flipping events, which is useful for applications such as analyzing reading behavior, optimizing user experiences, or gathering insights into content engagement.

Classification report (0 - notflipped, 1 - notflipped)

Model performance on the test data