Objects as context for detecting their semantic parts

TL;DR

This paper introduces a novel object-aware semantic part detection method that uses object appearance, class, and relative location modeling via OffsetNet to significantly improve detection accuracy.

Contribution

The paper proposes OffsetNet, a new network module that predicts variable part locations within objects, enhancing part detection by leveraging object context.

Findings

+5 mAP improvement on PASCAL-Part dataset

Outperforms previous methods on PASCAL-Part and CUB200-2011 datasets

Effective integration of object cues for semantic part detection

Abstract

We present a semantic part detection approach that effectively leverages object information.We use the object appearance and its class as indicators of what parts to expect. We also model the expected relative location of parts inside the objects based on their appearance. We achieve this with a new network module, called OffsetNet, that efficiently predicts a variable number of part locations within a given object. Our model incorporates all these cues to detect parts in the context of their objects. This leads to considerably higher performance for the challenging task of part detection compared to using part appearance alone (+5 mAP on the PASCAL-Part dataset). We also compare to other part detection methods on both PASCAL-Part and CUB200-2011 datasets.