Robotic strawberry harvesting requires machine vision system to have the ability to detect the presence, maturity, and location of strawberries. Strawberries, however, can easily be bruised, injured, and even damaged during robotic harvest if not picked properly because of their soft surfaces. Therefore, it is important to cut or pick the strawberry stems instead of picking the fruit directly. Additionally, real-time detection is critical for robotic strawberry harvesting to adapt to the changing field environment quickly. In this study, first, a detection algorithm was created for accurately localizing strawberries and their picking regions based on object detection network (YOLOv5s). The neck of YOLOv5s was replaced with a feature pyramid network (FPN) from path aggregation net (PA-Net) to reduce the complexity in network structure. This YOLOv5s model with FPN (YOLOv5s-FPN) was used to detect three maturity levels (immature, nearly mature, mature) of strawberries. Then, the model was used to detect picking region in strawberry stems using strawberry bounding boxes detected in the previous step as the input. For comparison, the original YOLOv5s was trained with same environment and datasets. The results showed that YOLOv5s-FPN model achieved the mean average precision (mAP) of 92.3% based on testing strawberry canopy dataset. In immature, nearly mature, and mature classes, it achieved an average precision of 93.6%, 91.7%, and 91.7%, respectively. For picking region, it achieved a mean average precision of 82.8%. Compared to YOLOv5s, the YOLOv5s-FPN had smaller size of 12.0 Mb (85.7% of YOLOv5s) and faster detection speed of 36.5ms (83.7% of YOLOv5s) on image of resolution 640×640 pixels. However, the performance YOlOv5s-FPN was equally good compared to YOLOv5s (mAP in strawberry detection: 92.5%; mAP in picking region: 82.6%). The YOLOv5s-FPN developed in this study showed good potential as a means for providing real-time detection of strawberry locations and corresponding stem regions for robotic twisting or cutting of stem as a way to harvest strawberries.