← 返回 Skills 市场

Image 3D Scene Reconstruction | 图像3D场景重建

Name: Image 3D Scene Reconstruction | 图像3D场景重建
Author: moroiser

作者 Morois · GitHub ↗ · v0.0.1 · MIT-0

cross-platform ✓ 安全检测通过

总下载

当前安装

版本数

在 OpenClaw 中安装

/install image-3d-scene-reconstruction

功能描述

Reconstruct 3D scenes from single images using depth estimation. 从单张图片重建3D场景结构（深度图、点云、Mesh）。

使用说明 (SKILL.md)

Image 3D Scene Reconstruction | 图像3D场景重建

从卫星图、航拍图或普通照片重建三维场景结构。基于 DA3Metric-Large（Depth Anything 3）深度估计模型，单张图片即可输出深度图、点云和 3D 模型。

Reconstruct 3D scenes from satellite, aerial, or regular photos. Based on DA3Metric-Large (Depth Anything 3), outputs depth maps, point clouds, and 3D models from a single image.

能力 | Capabilities

单图深度估计：输入一张图片，输出米制深度图（米为单位）
点云生成：从深度图反投影生成彩色 3D 点云
3DGS 输出：模型内置 3D Gaussian Splatting 能力
相机位姿估计：自动估计相机内外参
多图融合：支持多张图片输入做场景融合

使用方式 | Usage

快速开始

cd ~/.openclaw/workspace/projects/image-3d-scene-reconstruction
python3 scripts/reconstruct.py --input photo.jpg --output output/

Python API

from depth_anything_3.api import DepthAnything3
import cv2

model = DepthAnything3.from_pretrained('depth-anything/DA3Metric-Large')
model = model.cuda().eval()

img = cv2.imread('photo.jpg')
pred = model.inference([img])

depth = pred.depth[0]        # [H, W] 米制深度
extrinsics = pred.extrinsics  # 相机外参
intrinsics = pred.intrinsics  # 相机内参

CLI

# 单张图片 → 3D 输出
python3 -m depth_anything_3.cli image photo.jpg --export-dir output/ --export-format glb

# 多张图片 → 融合场景
python3 -m depth_anything_3.cli images ./photos/ --export-dir output/

依赖 | Dependencies

包	用途
depth-anything-3	深度估计 + 3D 重建引擎
opencv-python	图像处理
torch + torchvision	PyTorch 深度学习框架
open3d	点云处理（可选）
trimesh	Mesh 处理（可选）

硬件要求 | Hardware

GPU：NVIDIA GPU，6GB+ VRAM（GTX 1060 及以上）
CUDA：12.1+（PyTorch 2.5+）
CPU 模式：可用但极慢，仅推荐测试

项目文件 | Project Files

详见 ~/.openclaw/workspace/projects/image-3d-scene-reconstruction/README.md

安全使用建议

This skill appears benign as provided, but it is only a set of instructions. If you use it, review and trust the Python package, model download, and any local scripts before running them, especially because the skill itself does not include or pin those components.

功能分析

Type: OpenClaw Skill Name: image-3d-scene-reconstruction Version: 0.0.1 The skill bundle documentation describes a legitimate-sounding 3D scene reconstruction tool based on the Depth Anything 3 model. The provided files (_meta.json and SKILL.md) contain standard metadata, usage examples, and dependency lists (e.g., torch, opencv-python) consistent with a computer vision project, with no evidence of malicious intent, data exfiltration, or prompt injection.

能力评估

ℹ Purpose & Capability

The stated purpose is coherent with the documented depth-estimation, point-cloud, and 3D-model workflows, but the skill is documentation-only and references scripts/project files that are not included in the artifact set.

✓ Instruction Scope

The instructions are normal user-directed usage examples for image reconstruction and do not attempt to override the agent, force autonomous execution, or change stopping conditions.

ℹ Install Mechanism

There is no install spec, while SKILL.md lists Python dependencies and references a local script/module. This is a provenance and reproducibility note rather than evidence of malicious behavior.

ℹ Credentials

GPU/CUDA and ML dependencies are proportionate for 3D reconstruction, but they are not declared as required binaries or installed by the skill metadata.

✓ Persistence & Privilege

No credentials, privileged paths, persistence, background execution, or account mutation are declared or shown in the provided artifacts.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install image-3d-scene-reconstruction
安装完成后，直接呼叫该 Skill 的名称或使用 /image-3d-scene-reconstruction 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v0.0.1

1. Initial release. 首次发布。2. Support depth estimation, point cloud generation, 3DGS output. 支持深度估计、点云生成、3DGS 输出。

元数据

Slug image-3d-scene-reconstruction

版本 0.0.1

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题