LangSplat: 3D Language Gaussian Splatting
Authors:
Top Cited Papers
- 16 June 2024
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 20051-20060
Abstract
Humans live in a 3D world and commonly use natural language to interact with a 3D scene. Modeling a 3D language field to support open-ended language queries in 3D has gained increasing attention recently. This paper introduces LangSplat, which constructs a 3D language field that enables precise and efficient open-vocabulary querying within 3D spaces. Unlike existing methods that ground CLIP language embeddings in a NeRF model, LangSplat advances the field by utilizing a collection of 3D Gaussians, each encoding language features distilled from CLIP, to represent the language field. By employing a tile-based splatting technique for rendering language features, we circumvent the costly rendering process inherent in NeRF. Instead of directly learning CLIP embeddings, LangSplat first trains a scene-wise language autoencoder and then learns language features on the scene-specific latent space, thereby alleviating substantial memory demands imposed by explicit modeling. Existing methods struggle with imprecise and vague 3D language fields, which fail to discern clear boundaries between objects. We delve into this issue and propose to learn hierarchical semantics using SAM, thereby eliminating the need for extensively querying the language field across various scales and the regularization of DINO features. Extensive experimental results show that LangSplat significantly outperforms the previous state-of-the-art method LERF by a large margin. Notably, LangSplat is extremely efficient, achieving a 199 x speedup compared to LERF at the resolution of 1440 x 1080. We strongly recommend readers to check out our video results at https://langsplat.github.io/Keywords
All Related Versions
This publication has 25 references indexed in Scilit:
- Segment anything model for medical image analysis: An experimental studyMedical Image Analysis, 2023
- TensoRF: Tensorial Radiance Fields2022
- Learning Dynamic Facial Radiance Fields for Few-Shot Talking Head Synthesis2022
- Neural Feature Fusion Fields: 3D Distillation of Self-Supervised 2D Image Representations2022
- Learning to Prompt for Vision-Language ModelsInternational Journal of Computer Vision, 2022
- Direct Voxel Grid Optimization: Super-fast Convergence for Radiance Fields Reconstruction2022
- ScanQA: 3D Question Answering for Spatial Scene Understanding2022
- D-NeRF: Neural Radiance Fields for Dynamic Scenes2021
- pixelNeRF: Neural Radiance Fields from One or Few Images2021
- IQA: Visual Question Answering in Interactive Environments2018