DRCNN-WS: A Novel Approach for High-Resolution Video Using Recurrent Neural Networks and Walrus Search

• Published in: Proceedings (International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions) pp. 1 - 6
• Main Authors: NJ, Satheesh Kumar, CH, Arun
• Format: Conference Proceeding
• Language: English
• Published: IEEE 14.03.2024
• Subjects: Compute Unified Device Architecture; Deep Recurrent Neural Network; Energy measurement; High-efficiency video coding; Higher resolution video; Measurement uncertainty; Noise reduction; Recurrent neural networks; Streaming media; Training; Video coding; Walrus optimizer
• ISSN: 2769-2884
• DOI: 10.1109/ICRITO61523.2024.10522118

In general, a video that contains a greater number of pixels is specified as the higher-resolution video. Consequently, greater comprehensive and clear images could be obtained. Through the number of pixels, the resolution is measured. Among several domains, including marketing, entertainment, education, modern communication, and so on, high-quality video has a significant role. In this context, a lot of approaches are introduced in order to get a high-resolution video. These methods attempt to acquire a high-resolution video, but have difficulties during training and testing which is more time-consuming. This work employs a Deep Recurrent Neural Network-Walrus Search (DRCNN-WS) algorithm that enables the RCNN and an optimization algorithm for high-quality video. The input video for the developed DRCNN-WS approach is attained through the UCF101 Dataset which holds a larger amount of video clips. Once the video is collected, Video Denoising, Frame Rate Conversion, and Video Stabilization pre-processing steps are added. These steps eliminate the irrelevant data in the input. The DRCNN-WS model accomplishes the objective of getting a higher-resolution video, while the Walrus Search (which includes the initial search strategy) is exploited in order to tune the parameters of the developed model. The experimental results for the purpose of the evaluation of the performance are conducted, and the parameter measures such as End-to-end delay latency, Error Rate, PSNR, energy efficiency, SSIM, and Packet loss rate measures are exploited. DRCNN-WS model delivers high-resolution video with higher performance rates.