MALE FERTILITY DETECTION USING DETECTRON2 & CSR-DCF

Abstract: This paper introduces an optimized approach for detecting and tracking sperm in phase-contrast microscopy image sequences, with the aim of improving fertility analysis and sperm behavior studies. The proposed method integrates advanced object detection techniques with a modified multi- object tracking algorithm to achieve superior accuracy and robustness. Through comprehensive experimentation, our approach demonstrates exceptional performance in challenging scenarios such as high-density sperm samples, occlusions, and collisions, achieving an F1 score of 96.61% in tracking accuracy. This optimized algorithm holds significant promise for advancing research in reproductive health.