TransRefer3D: Entity-and-Relation Aware Transformer for Fine-Grained 3D Visual Grounding

Abstract

Recently proposed fine-grained 3D visual grounding is an essential and challenging task, whose goal is to identify the 3D object referred by a natural language sentence from other distractive objects of the same category. Existing works usually adopt dynamic graph networks to indirectly model the intra/inter-modal interactions, making the model difficult to distinguish the referred object from distractors due to the monolithic representations of visual and linguistic contents. In this work, we exploit Transformer for its natural suitability on permutation-invariant 3D point clouds data and propose a TransRefer3D network to extract entity-and-relation aware multimodal context among objects for more discriminative feature learning. Concretely, we devise an Entity-aware Attention (EA) module and a Relation-aware Attention (RA) module to conduct fine-grained cross-modal feature matching. Facilitated by co-attention operation, our EA module matches visual entity features with linguistic entity features while RA module matches pair-wise visual relation features with linguistic relation features, respectively. We further integrate EA and RA modules into an Entity-and-Relation aware Contextual Block (ERCB) and stack several ERCBs to form our TransRefer3D for hierarchical multimodal context modeling. Extensive experiments on both Nr3D and Sr3D datasets demonstrate that our proposed model significantly outperforms existing approaches by up to 10.6% and claims the new state-of-the-art. To the best of our knowledge, this is the first work investigating Transformer architecture for fine-grained 3D visual grounding task.

arXiv commentACM MM2021

References in corpus (13)

• Adam: A Method for Stochastic Optimization
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• Hierarchical Question-Image Co-Attention for Visual Question Answering
• Beyond Bilinear: Generalized Multimodal Factorized High-order Pooling for Visual Question Answering
• ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes
• Pre-Trained Image Processing Transformer
• Deep Modular Co-Attention Networks for Visual Question Answering
• Attention on Attention for Image Captioning
• Visual Relationship Detection with Visual-Linguistic Knowledge from Multimodal Representations
• Understanding the Difficulty of Training Transformers
• Visual Semantic Reasoning for Image-Text Matching
• Graph Structured Network for Image-Text Matching
• Referring Image Segmentation via Cross-Modal Progressive Comprehension

Cited by in corpus (2)

• 3D-SPS: Single-Stage 3D Visual Grounding via Referred Point Progressive Selection
• EDA: Explicit Text-Decoupling and Dense Alignment for 3D Visual Grounding