Abstract: packaging，cigarettes，etc.) holds significant importance. Traditional detection methods often struggle to accurately identify minor defects and are relatively slow in detection speed. This paper proposes an improved detection method that combines the CBAM with a small object detection layer. By enhancing feature extraction capabilities and improving the recognition accuracy of minor defects， the detection performance is effectively improved. In experiments targeting the detection of surface defects on cigarette-like rod-shaped objects，the improved model exhibits excellent performance in recognizing minor defects，achieving an mAP@0.5 of 0.972. This model strikes a good balance between detection accuracy，lightweight，and detection speed. Experiments show that this method can be effectively applied to the appearance quality inspection of cigarette-like rod-shaped objects，such as appearance flaws and cracks， contributing to enhancing the efficiency and reliability of quality control in the cigarette manufacturing industry.