REVIEW ON SSP-NET: SCALABLE SEQUENTIAL PYRAMID NETWORKS
International Research Journal of Modernization in Engineering Technology and Science · 2024
In this study, a highly scalable convolutional neural network for real-time 3D human posture regression from RGB photos is introduced: The Scalable Sequential Pyramid Networks (SSP-Net).SSP-Net demonstrates its capability at 120 frames per second and makes reasonable predictions at more than 200 frames per second, achieving amazing prediction speeds.Due to the method's invariance to feature map size, low-resolution feature maps can produce competitive results and multi-resolution intermediate supervisions.Comprehensive tests on the MPI-INF-3DHP and Human3.6Mdatasets demonstrate SSP-Net's efficacy and accuracy.This survey article provides important insights on SSP-Net's contributions to several sectors by thoroughly examining its architecture, components, applications, and comparative characteristics.The survey explores the sophisticated feature extraction algorithms, sequential scalability, and distinct pyramid topologies of SSP-Net.A detailed examination of the main elements, uses, and a comparison with other scalable architectures offers a sophisticated grasp of the benefits, drawbacks, and possibilities of SSP-Net.Case studies highlight SSP-Net's adaptability and effectiveness in a variety of scenarios by illuminating its real-world effects.The survey adds to the current discussion on SSP-Net by addressing issues and outlining potential avenues for future study.It is an invaluable tool for scholars, practitioners, and enthusiasts who are interested in the latest advancements in scalable pyramid structures in neural networks.