Medical imaging is one of the critical tasks in the image segmentation process. Knee detection comes under it. It is done by using U-Net segmentation. This is a very powerful deep learning approach that has been effective in the medical field for a very long time including knee detection. This U-Net segmentation has areas to improve which lead to derivation of various techniques from the main algorithm like U-Net++,pspnet,Aunet and many more. For finding the efficiency and accuracy for medical imaging we choose three different algorithms along with the U-Net segmentation to compare the factors like Number of parameters, FLOPS,Memory usage and speed. This comparison leads to conclusions finding which is the most efficient model in terms of both memory usage and computation resources. The method used for the comparison is training the models and summarizing the results. From the summarization we take the params to compare and find the result. By the process of the comparison we can conclude that the PSPNet model achieved high accuracy. Whereas, U-Net or U-Net + + achieved higher computational speed. For the proposed medical image segmentation application, PspNet achieved 85.4% results.
No takes yet. Share an insight, caveat, or question.
venkatesh et al. (2024) studied this question.