Abstract: Currently, the production of image segmentation semantic label datasets relies on manual annotation, which is labor-intensive and inefficient. We propose a fast image semantic labeling method for common objects to address these challenges based on the visual background extractor (ViBe) algorithm. To mitigate the issue of foreground point pixel updates over time before processing the keyframes, we introduce an event-triggered background update mechanism, effectively reducing the occurrence of voids in the foreground target. First, we collect and preprocess video sequences for experiments. We apply the ViBe algorithm to detect dynamic objects in images and perform foreground segmentation. Then, we extract keyframes of target objects from the image sequences using a sliding window algorithm. Next, we binarize the extracted keyframes, apply image filtering, and perform label pixel classification and filling to generate labeled images. Experimental results demonstrate that this method substantially simplifies the production process of semantic label datasets. Notably, the optimized algorithm achieves high label coverage accuracy, with an average coverage accuracy of 97.18%, while maintaining an average processing time of 96.10 ms per frame.