Pointcept
/

Utonia

@@ -15,6 +15,7 @@ This repository contains the model weights for **Utonia**, a step toward one-fro
 - **Paper:** [Utonia: Toward One Encoder for All Point Clouds](https://arxiv.org/abs/2603.03283)
 - **Project Page:** [https://pointcept.github.io/Utonia/](https://pointcept.github.io/Utonia/)
 - **Inference:** [https://github.com/Pointcept/Utonia](https://github.com/Pointcept/Utonia)
 ## Models
 The default model take `[coord, color, normal]` as input, which can deal with input without color and normal. If colors and normals are not available, please set them to zeros.
@@ -27,7 +28,7 @@ The default model take `[coord, color, normal]` as input, which can deal with in
 We dream of a future where point clouds from all domains can come together to shape a single model that benefits them all. Toward this goal, we present Utonia, a first step toward training a single self-supervised point transformer encoder across diverse domains, spanning remote sensing, outdoor LiDAR, indoor RGB-D sequences, object-centric CAD models, and point clouds lifted from RGB-only videos. Despite their distinct sensing geometries, densities, and priors, Utonia learns a consistent representation space that transfers across domains. This unification improves perception capability while revealing intriguing emergent behaviors that arise only when domains are trained jointly. Beyond perception, we observe that Utonia representations can also benefit embodied and multimodal reasoning: conditioning vision-language-action policies on Utonia features improves robotic manipulation, and integrating them into vision-language models yields gains on spatial reasoning. We hope Utonia can serve as a step toward foundation models for sparse 3D data, and support downstream applications in AR/VR, robotics, and autonomous driving.
 ## Usage
-For detailed installation, data preparation, and testing instructions, please refer to the [inference demo](https://github.com/Pointcept/Utonia).
 ## Citation
 If you find Utonia useful in your research, please cite the following papers:

 - **Paper:** [Utonia: Toward One Encoder for All Point Clouds](https://arxiv.org/abs/2603.03283)
 - **Project Page:** [https://pointcept.github.io/Utonia/](https://pointcept.github.io/Utonia/)
 - **Inference:** [https://github.com/Pointcept/Utonia](https://github.com/Pointcept/Utonia)
+- **Pretrain:** [https://github.com/Pointcept/Pointcept](https://github.com/Pointcept/Pointcept)
 ## Models
 The default model take `[coord, color, normal]` as input, which can deal with input without color and normal. If colors and normals are not available, please set them to zeros.
 We dream of a future where point clouds from all domains can come together to shape a single model that benefits them all. Toward this goal, we present Utonia, a first step toward training a single self-supervised point transformer encoder across diverse domains, spanning remote sensing, outdoor LiDAR, indoor RGB-D sequences, object-centric CAD models, and point clouds lifted from RGB-only videos. Despite their distinct sensing geometries, densities, and priors, Utonia learns a consistent representation space that transfers across domains. This unification improves perception capability while revealing intriguing emergent behaviors that arise only when domains are trained jointly. Beyond perception, we observe that Utonia representations can also benefit embodied and multimodal reasoning: conditioning vision-language-action policies on Utonia features improves robotic manipulation, and integrating them into vision-language models yields gains on spatial reasoning. We hope Utonia can serve as a step toward foundation models for sparse 3D data, and support downstream applications in AR/VR, robotics, and autonomous driving.
 ## Usage
+For detailed installation, data preparation, and testing instructions, please refer to the [inference demo](https://github.com/Pointcept/Utonia). The `utonia.pth` is the simplified version for [inference](https://github.com/Pointcept/Utonia), while `pretrain-utonia-v1m1-0-base_stagev2.pth` is the original weight from pretraining and can be evaluated in [Pointcept](https://github.com/Pointcept/Pointcept).
 ## Citation
 If you find Utonia useful in your research, please cite the following papers: