MetaAnchor: Learning to Detect Objects with Customized Anchors

Understand

We propose a novel and flexible anchor mechanism named MetaAnchor for object detection frameworks.

• Unlike many previous detectors model anchors via a predefined manner, in MetaAnchor anchor functions could be dynamically generated from the arbitrary customized prior boxes.
• Taking advantage of weight prediction, MetaAnchor is able to work with most of the anchor-based object detection systems such as RetinaNet.
• Compared with the predefined anchor scheme, we empirically find that MetaAnchor is more robust to anchor settings and bounding box distributions; in addition, it also shows the potential on transfer tasks.