Learning Equivariant Segmentation with Instance-Unique Querying
This work addresses the challenge of learning more discriminative and robust queries for instance segmentation, which is important for computer vision applications, but it is incremental as it builds on existing query-based models.
The paper tackles the problem of improving instance segmentation in query-based models by introducing a training framework that enforces dataset-level uniqueness and transformation equivariance in query embeddings, resulting in performance gains of 1.6-3.2 AP on COCO and 2.7 AP on LVISv1.
Prevalent state-of-the-art instance segmentation methods fall into a query-based scheme, in which instance masks are derived by querying the image feature using a set of instance-aware embeddings. In this work, we devise a new training framework that boosts query-based models through discriminative query embedding learning. It explores two essential properties, namely dataset-level uniqueness and transformation equivariance, of the relation between queries and instances. First, our algorithm uses the queries to retrieve the corresponding instances from the whole training dataset, instead of only searching within individual scenes. As querying instances across scenes is more challenging, the segmenters are forced to learn more discriminative queries for effective instance separation. Second, our algorithm encourages both image (instance) representations and queries to be equivariant against geometric transformations, leading to more robust, instance-query matching. On top of four famous, query-based models ($i.e.,$ CondInst, SOLOv2, SOTR, and Mask2Former), our training algorithm provides significant performance gains ($e.g.,$ +1.6 - 3.2 AP) on COCO dataset. In addition, our algorithm promotes the performance of SOLOv2 by 2.7 AP, on LVISv1 dataset.