Abstract: To address the impact of common challenges such as small targets and dense arrangement in the 3C electronic component sorting scenario on the performance of multi-object tracking, proposes a lightweight multi-object tracking method based on improved YOLOv8n and ByteTrack. In the object detection stage, aiming at the difficulties in detecting small targets and mutual occlusion between targets, the original neck part of the YOLOv8n network is replaced with the Gold-YOLO module to enhance multi-scale feature fusion capability, and the ACmix attention mechanism is embedded in the N3 output layer of Gold-YOLO to strengthen the feature extraction ability for small and occluded targets. In the object tracking stage, algorithm parameters are optimized and a nested box filtering algorithm is added. It effectively suppresses mismatches caused by overlapping detection boxes and improves tracking robustness. The improved model's performance was verified using a self-built dataset. The experimental results show that the proposed improved YOLOv8n object detection algorithm and improved ByteTrack object tracking algorithm perform excellently in the electronic component sorting scenario. On the self-built dataset, the method achieved mAP@50 of 91. 32% in the object detection. In the object tracking, it can achieve the MOTA of 86. 72%, IDF1 of 93. 13%, number of identity switches (IDSW) of 0 times, and frame rate of 27. 65 fps. The proposed algorithm takes into account detection accuracy, tracking robustness and real-time performance in complex industrial scenarios, providing an efficient and reliable target tracking solution for industrial intelligent sorting scenarios.