Abstract 3D model-based object recognition has been a noticeable research trend in recent years. Common methods ?nd 2D-to-3D correspondences and make recognition decisions by pose estimation, whose ef?ciency usually suffers from noisy correspondences caused by the increasing number of target objects. To overcome this scalability bottleneck, we propose an ef?cient 2D-to-3D correspondence ?ltering approach, which combines a light-weight neighborhoodbased step with a ?ner-grained pairwise step to remove spurious correspondences based on 2D/3D geometric cues. On a dataset of 300 3D objects, our solution achieves ?10 times speed improvement over the baseline, with a comparable recognition accuracy. A parallel implementation on a quad-core CPU can run at ?3fps for 1280×720 images.