SGRec3D: Self-Supervised 3D Scene Graph Learning via Object-Level Scene Reconstruction
This addresses the problem of limited labeled data for 3D scene graph learning in computer vision, enabling use of large-scale datasets, though it is incremental as it builds on existing pre-training approaches.
The paper tackles the challenge of learning 3D scene graphs without requiring relationship labels by introducing SGRec3D, a self-supervised pre-training method that reconstructs 3D scenes from a graph bottleneck, resulting in state-of-the-art performance with +10% improvement on object prediction and +4% on relationship prediction.
In the field of 3D scene understanding, 3D scene graphs have emerged as a new scene representation that combines geometric and semantic information about objects and their relationships. However, learning semantic 3D scene graphs in a fully supervised manner is inherently difficult as it requires not only object-level annotations but also relationship labels. While pre-training approaches have helped to boost the performance of many methods in various fields, pre-training for 3D scene graph prediction has received little attention. Furthermore, we find in this paper that classical contrastive point cloud-based pre-training approaches are ineffective for 3D scene graph learning. To this end, we present SGRec3D, a novel self-supervised pre-training method for 3D scene graph prediction. We propose to reconstruct the 3D input scene from a graph bottleneck as a pretext task. Pre-training SGRec3D does not require object relationship labels, making it possible to exploit large-scale 3D scene understanding datasets, which were off-limits for 3D scene graph learning before. Our experiments demonstrate that in contrast to recent point cloud-based pre-training approaches, our proposed pre-training improves the 3D scene graph prediction considerably, which results in SOTA performance, outperforming other 3D scene graph models by +10% on object prediction and +4% on relationship prediction. Additionally, we show that only using a small subset of 10% labeled data during fine-tuning is sufficient to outperform the same model without pre-training.