Calculus-based image processing on GPU using python with Pytorch

Moreover, visualization is a powerful way to learn due to the appreciation of the visual stimuli. Thus, we provide graphic results of the various terms found in multivariable calculus applied to images rather than the conventional applications found in physics and engineering textbooks. We believe that images provide a more intuitive feel when observed. This makes it easier to understand, appreciate and enjoy several aspects of multivariable calculus and partial differential equations without being bogged down by the mathematics. The terms such as gradient, norm, vector, divergence, curvature, eigenvectors should be familiar and even if not, the visual representations hopefully, should create a lasting impression on the reader as they study the concepts. This will encourage hands-on learning and understanding of these formulations and how they feature in image processing and computer vision. Additionally, the Pytorch implementations of the algorithms exploiting the CUDA framework on NVIDIA GPUs are presented for faster operation.