This study proposes an improved YOLOv8+ model combined with image processing methods to accurately and comprehensively identify the ripeness of strawberries in complex environments. The ECA attention mechanism is added to the backbone network of YOLOv8+ to enhance feature extraction and improve model performance. The Focal-EIIOU loss function is used to address the imbalance between easy and difficult-to-classify samples. The centerline of ripe strawberries is extracted, and the red pixels in this centerline are counted using the HSV color space to quantify ripeness. The method classifies ripe strawberries into fully ripe and not fully ripe categories. The results show that the improved YOLOv8+ model achieves an accuracy of 97.81%, a recall of 96.36%, and an F1 score of 97.07%. The image processing method for classifying ripe strawberries has an accuracy of 91.91%, a false positive rate of 5.03%, and a false negative rate of 14.28%. The study demonstrates the effectiveness of the proposed method in real-time detection and classification of strawberry ripeness, which can guide selective harvesting by fruit-picking robots.This study proposes an improved YOLOv8+ model combined with image processing methods to accurately and comprehensively identify the ripeness of strawberries in complex environments. The ECA attention mechanism is added to the backbone network of YOLOv8+ to enhance feature extraction and improve model performance. The Focal-EIIOU loss function is used to address the imbalance between easy and difficult-to-classify samples. The centerline of ripe strawberries is extracted, and the red pixels in this centerline are counted using the HSV color space to quantify ripeness. The method classifies ripe strawberries into fully ripe and not fully ripe categories. The results show that the improved YOLOv8+ model achieves an accuracy of 97.81%, a recall of 96.36%, and an F1 score of 97.07%. The image processing method for classifying ripe strawberries has an accuracy of 91.91%, a false positive rate of 5.03%, and a false negative rate of 14.28%. The study demonstrates the effectiveness of the proposed method in real-time detection and classification of strawberry ripeness, which can guide selective harvesting by fruit-picking robots.