Skip to content
This repository has been archived by the owner on Aug 16, 2024. It is now read-only.

Latest commit

 

History

History
46 lines (16 loc) · 1.2 KB

README.md

File metadata and controls

46 lines (16 loc) · 1.2 KB
Raw

Humanoid at Home Depth Estimation

Demos

House_video_depth.mov

Depth Estimation

3d_scene.mp4

3D Scene Reconstruction

pc_demo.mov
Screen.Recording.2024-07-03.at.10.49.29.AM.mov

Instructions

  1. run_video_depth_only.py: Get depth estimation inference frames from video

  2. depth_to_pc.py: Get point clouds for each frame in video, by cross referencing original (*_OG.png) and generated depth frames. Change camera intrinsics appropriately to generate accurate point clouds.

  3. video_pc_testing.py: To test camera view ports, angles, and positions to generate a point cloud video

  4. video_pc_actual.py: generate point cloud video, camera rotates on x-y plane in a circle for 3d effects. Utilizes parameters found after video_pc_testing.py.

  5. video_pc_mesh.py: generates mesh from pointcloud, using KDTree for normal estimation and Poisson Surface Reconstruction for creating smooth, detailed surfaces

Requirements

environment.yml