3D CoCa: Contrastive Learners are 3D Captioners
It addresses the problem of generating natural language descriptions for 3D scenes, which is important for applications like robotics and augmented reality, by introducing a novel joint training approach that eliminates the need for external detectors.
The paper tackles the challenge of 3D captioning by proposing 3D CoCa, a unified framework that combines contrastive learning and caption generation, achieving state-of-the-art performance with improvements of 10.2% and 5.76% in CIDEr on benchmarks.
3D captioning, which aims to describe the content of 3D scenes in natural language, remains highly challenging due to the inherent sparsity of point clouds and weak cross-modal alignment in existing methods. To address these challenges, we propose 3D CoCa, a novel unified framework that seamlessly combines contrastive vision-language learning with 3D caption generation in a single architecture. Our approach leverages a frozen CLIP vision-language backbone to provide rich semantic priors, a spatially-aware 3D scene encoder to capture geometric context, and a multi-modal decoder to generate descriptive captions. Unlike prior two-stage methods that rely on explicit object proposals, 3D CoCa jointly optimizes contrastive and captioning objectives in a shared feature space, eliminating the need for external detectors or handcrafted proposals. This joint training paradigm yields stronger spatial reasoning and richer semantic grounding by aligning 3D and textual representations. Extensive experiments on the ScanRefer and Nr3D benchmarks demonstrate that 3D CoCa significantly outperforms current state-of-the-arts by 10.2% and 5.76% in CIDEr at 0.5IoU, respectively. Code will be available at https://github.com/AIGeeksGroup/3DCoCa.