This project focuses on the analysis of corner separation phenomena in axial compressors using computational fluid dynamics (CFD) data. The primary objective is to develop a Python-based methodology for detecting critical points and classifying them as saddles, nodes, or foci in the near-wall velocity field of a high-fidelity large-eddy simulation (LES) dataset. Understanding the topology of these flow structures is essential for evaluating secondary flow losses and separation-driven performance degradation in turbomachinery. The dataset used originates from a numerical simulation by Ventosa-Molina et al. (2022), featuring a compressor cascade with and without rotational (Coriolis) forces. Post-processing was conducted using ParaView for preliminary visualization, followed by a detailed analysis in Python. Cartesian velocity vectors were converted into cylindrical coordinates to better align with the curvature of the domain near the hub. The detection algorithm identifies critical points by evaluating wall-parallel velocity gradients, and classifies them using the eigenvalues of the local Jacobian matrix. A comparative study was carried out for both rotating and non-rotating configurations, under instantaneous and time-averaged conditions. The results show that the location and type of critical points vary with the threshold used for detection, the inclusion of Coriolis effects, and whether the flow field is instantaneous or averaged. In the non-rotating case, a typical open separation pattern was observed at low thresholds, while higher thresholds revealed extended networks of critical points near the suction surface. In the rotating case, the separation was less pronounced, with critical points clustering near the blade surface and turbulent regions showing signs of suppression. The methodology proves effective in reproducing the expected flow topology patterns documented in the literature, such as those described by Dawkins and Gbadebo. However, numerical limitations, machine precision, and the need for threshold-based gradient evaluation constrain the accuracy and robustness of the method. Future improvements may involve extending the analysis to three-dimensional streamline tracing, incorporating unsteady flow tracking, or validating results experimentally. These findings provide further insight into the structure of corner separation and highlight the importance of critical point analysis in compressor design and optimization.