<p>To achieve effective sorting by courier robots, object detection plays a critical role. Given the challenges posed by complex backgrounds, small detection targets, and varying tilt angles in courier images, this paper proposes an enhanced YOLOv8 object detection method. First, the CBAM attention mechanism is introduced to improve the model’s focus on small targets. Next, BiFPN and SE modules are employed for feature fusion to enhance the representation of object features. To handle tilted objects, a rotation detection head is added to the regression branch, and the KFIoU loss function is integrated to improve positional accuracy. Finally, the detected regions are extracted, and Tesseract-OCR is used for object recognition. Experimental results on the dataset demonstrate that the improved YOLOv8 model improves by 11.1%, 9.9%, and 4.9% in precision, recall, and mAP, respectively, compared to the original YOLOv8 model. Additionally, model parameters and computational load are reduced. The enhanced model exhibits improved accuracy and robustness, effectively recognizing recipient information on courier labels. This method shows potential for application in real-world courier sorting scenarios.</p>