A Virtual-Real Interaction Approach to Object Instance Segmentation in Traffic Scenes

摘要

Object instance segmentation in traffic scenes is an important research topic. For training instance segmentation models, synthetic data can potentially complement real data, alleviating manual effort on annotating real images. However, the data distribution discrepancy between synthetic data and real data hampers the wide applications of synthetic data. In light of that, we propose a virtual-real interaction method for object instance segmentation. This method works over synthetic images with accurate annotations and real images without any labels. The virtual-real interaction guides the model to learn useful information from synthetic data while keeping consistent with real data. We first analyze the data distribution discrepancy from a probabilistic perspective, and divide it into image-level and instance-level discrepancies. Then, we design two components to align these discrepancies, i.e., global-level alignment and local-level alignment. Furthermore, a consistency alignment component is proposed to encourage the consistency between the global-level and the local-level alignment components. We evaluate the proposed approach on the real Cityscapes dataset by adapting from virtual SYNTHIA, Virtual KITTI, and VIPER datasets. The experimental results demonstrate that it achieves significantly better performance than state-of-the-art methods.