This study proposes a method for strawberry detection and ripeness classification using the YOLOv8+ model combined with image processing. The YOLOv8+ model is enhanced with an ECA attention mechanism and Focal-EIOU loss to improve detection performance. The ECA mechanism helps capture long-distance dependencies in the image, while the Focal-EIOU loss addresses the imbalance between easy and difficult-to-classify samples. The model achieves high accuracy (97.81%), recall (96.36%), and F1 score (97.07) in identifying ripe and unripe strawberries. For ripeness classification, the centerline of ripe strawberries is extracted, and the red pixel ratio in the centerline is calculated as a new parameter. This ratio is used to classify strawberries as fully ripe or not fully ripe. The image processing method achieves an accuracy of 91.91%, with a false positive rate (FPR) of 5.03% and a false negative rate (FNR) of 14.28%. The study demonstrates that the improved YOLOv8+ model can accurately and comprehensively identify strawberry ripeness in complex environments, including varying lighting and occlusion conditions. The combination of deep learning and image processing enables efficient and accurate classification of strawberry ripeness, which can guide selective harvesting by fruit-picking robots. The results show that the proposed method is effective in real-time detection and classification of strawberries at different ripeness stages.This study proposes a method for strawberry detection and ripeness classification using the YOLOv8+ model combined with image processing. The YOLOv8+ model is enhanced with an ECA attention mechanism and Focal-EIOU loss to improve detection performance. The ECA mechanism helps capture long-distance dependencies in the image, while the Focal-EIOU loss addresses the imbalance between easy and difficult-to-classify samples. The model achieves high accuracy (97.81%), recall (96.36%), and F1 score (97.07) in identifying ripe and unripe strawberries. For ripeness classification, the centerline of ripe strawberries is extracted, and the red pixel ratio in the centerline is calculated as a new parameter. This ratio is used to classify strawberries as fully ripe or not fully ripe. The image processing method achieves an accuracy of 91.91%, with a false positive rate (FPR) of 5.03% and a false negative rate (FNR) of 14.28%. The study demonstrates that the improved YOLOv8+ model can accurately and comprehensively identify strawberry ripeness in complex environments, including varying lighting and occlusion conditions. The combination of deep learning and image processing enables efficient and accurate classification of strawberry ripeness, which can guide selective harvesting by fruit-picking robots. The results show that the proposed method is effective in real-time detection and classification of strawberries at different ripeness stages.