Detection of Objects in High-Definition Videos for Disaster Management
Gowri Shankar J, A. Suresh Kumar, Ramesh Sekaran, Manikandan Parasuraman, S. Annamalai, T Narmadha · 2023
This paper presents a object detection quality assessment (QA) algorithm for ultra-high definition (UHD) videos that is both efficient and precise. The proposed method is based on the use of a single-stream deep learning model, which allows for the automatic extraction of spatio-temporal information from video clips. It is made up of two subcomponents: a network for extracting spatial distortion features and a recurrent network for modifying the quality score in the continuous time dimension, both of which depend on long short-term memory (LSTM). When applied to databases of UHD-quality, extensive testing demonstrates that the recommended model performs better than alternative BVQA methods.