<p>Latest research explored into how Deep Learning (DL) can be used to accelerate topology optimization while reducing processing costs. Topology optimization, relying on finite element analysis (FEA) and iterative solvers, typically incurs substantial computational overheads, especially in evaluating intricate designs. This paper investigates the effectiveness of various advanced neural network models in achieving faster topology optimization. We evaluate and compare three such models using three well-structured datasets. Our method produces topology-optimized designs in both the two- and three-dimensional domains, illustrating the efficacy of DL in this domain. According to the findings, both the upgraded U-Net and Res-U-Net systems perform well as dependable DL strategies and dependable as faster topology optimization. Notably, the analysis reveals that Res-U-Net exhibits superior performance at higher iterations compared to U-Net. Furthermore, we show that, while preserving excellent accuracy, our suggested CNN method has a large time advantage over current state-of-the-art techniques for training.</p>