The use of additive manufacturing has increased significantly in recent years, particularly in the aerospace industry. However, AM materials often exhibit defects that adversely impact fatigue performance. This study examines the geometric and morphological features of critical defects observed in Ti-6Al-4V specimens. A framework for automatic fatigue failure analysis through computer vision is proposed. An AI-based tool was trained to identify critical defects, measure their proximity to the surface, and quantify 14 geometric and morphological features. The findings indicate that surface proximity is the most influential factor in fatigue life classification, with defects near the surface exerting a negative impact on performance. No clear trend was observed in defect morphology beyond a certain surface distance. For lack-of-fusion defects classified as critical, the X-parameter model was applied and a correlation of R 2 = 0 . 9 1 with the measured CTF was obtained.