Skip to content

Dataset · sbonilla

Stereo endoscopic videos for surgical 3D vision

sbonilla/imed

Recordings from two stereo endoscopes (~48 GB) for depth estimation, pose, and 3D reconstruction in surgery.

What's inside

The full version 2.0 release contains 341 video sequences across 43 scenes from seven sessions with a total volume of about 48 GB. Recordings were made using two stereo endoscopes on pig tissues (ex-vivo, post-mortem, and live in-vivo).

What you can do with it

  • Training depth estimation and camera pose models in endoscopy
  • Developing feature matching and novel view synthesis algorithms for deformable tissues
Tasks
Depth estimation
Size
up to 1K rows
Downloads this month
924
Likes
3

How to load it

from datasets import load_dataset

ds = load_dataset("sbonilla/imed")