Abstract: Aiming at the problem of high missed detection rate and poor real-time performance of stacked occluded workpieces in actual production scenarios, an improved YOLOv7 lightweight target detection method is proposed. The three feature layers obtained in the backbone network are integrated into the ECA attention mechanism respectively, which can better notice the occluded workpiece information and reduce the missed detection rate. The ordinary convolution of the backbone network and the head network is replaced by a deep separable convolution to reduce the number of parameters of the network model and improve the detection speed. An ELAN-DE module is proposed, which is associated with peripheral feature information. The depth separable convolution is introduced and the ECA attention mechanism is added to improve the stability of the system. The experimental results show that compared with the original algorithm, the improved YOLOv7 algorithm improves the average accuracy (mAP) by nearly 5.52%, the model size is reduced by nearly 21%, the detection speed is increased by 18.7 f/s, and the detection effect is also better than other classical target detection network algorithms, which meets the workpiece recognition requirements in the actual scene.