# Contextual Translation Embedding for Visual Relationship Detection and Scene Graph Generation

## TL;DR

This paper introduces a context-augmented translation embedding model for visual relationship detection and scene graph generation, improving recognition of both seen and unseen relations by incorporating contextual information.

## Contribution

It extends the VTransE model by integrating contextual features from the union of subject and object bounding boxes, enhancing relation recognition and generalization.

## Key findings

- Outperforms previous translation-based models on multiple benchmarks.
- Achieves state-of-the-art results on unseen relation detection.
- Provides promising results for scene graph generation.

## Abstract

Relations amongst entities play a central role in image understanding. Due to the complexity of modeling (subject, predicate, object) relation triplets, it is crucial to develop a method that can not only recognize seen relations, but also generalize to unseen cases. Inspired by a previously proposed visual translation embedding model, or VTransE, we propose a context-augmented translation embedding model that can capture both common and rare relations. The previous VTransE model maps entities and predicates into a low-dimensional embedding vector space where the predicate is interpreted as a translation vector between the embedded features of the bounding box regions of the subject and the object. Our model additionally incorporates the contextual information captured by the bounding box of the union of the subject and the object, and learns the embeddings guided by the constraint predicate $\approx$ union (subject, object) $-$ subject $-$ object. In a comprehensive evaluation on multiple challenging benchmarks, our approach outperforms previous translation-based models and comes close to or exceeds the state of the art across a range of settings, from small-scale to large-scale datasets, from common to previously unseen relations. It also achieves promising results for the recently introduced task of scene graph generation.