Given the different sizes of sponge strips and small color difference with the background in the assembly environment of automobile door panels, an detection model based on improved YOLO11n is put forward. This study proposes a C3k2_IDWC module to optimize the standard convolution via the multi-branch feature extraction mechanism to enhance the multi-scale feature extraction ability of the model. Meanwhile, the DSWTHead detection head is proposed to employ wavelet transform convolution to extract the global structural features and detailed texture information. The point-by-point convolution is adopted to integrate inter-channel information, optimize the modeling of global and detailed information, and enhance the context information of the detection head. Additionally, the ADown module is introduced to further improve the ability of global information modeling and feature expression of the model. Experimental results show that compared with the original YOLO11n model, the improved model features sound improvements in the accuracy, recall, mAP@0.5, and mAP@0.5:0.95, an increase of 9.3%, 18.1%, 11.6%, and 18.6% respectively. The number of parameters and calculations is reduced, thus improving the detection accuracy of sponge strips for automobile door panels.