Sickle Cell Disease SCD (SCD) is a genetic hemoglobinopathy characterized by the production of abnormal hemoglobin, leading to the destruction of erythrocytes and chronic organ damage. This study addresses several limitations observed in previous studies by introducing an integrated approach that combines cell identification and detection into a unified framework. The primary objective here is to provide quantitative measures of different types of red blood cells, which are fundamentally important in a more holistic diagnosis than mere classification. In this paper, the performance of three state-of-the-art single-stage object detectors, YOLOv4, YOLOv5, and YOLOv8n, in the task of detection of sickle cells using deep learning is analyzed. The results indicated that YOLOv8n performs better than the other two models, and achieved an important score of 96.1% mAP at an IoU of 50%. Yolov8n uses advanced IoU and non-maximum suppression algorithms, which leads to accurate detections, less overlap, and eventually contributes to improved accuracy. Architectural improvements, such as a separated head and improved feature extractor, likely contributed to YOLOv8n outperforming its precursors. Results outline the importance of YOLOv8n as a rapidly emerging tool that can be deployed in automated detection and evaluation systems for SCD thereby enabling more accurate diagnosis.