Meshroom in Google Colab (cloud)

You don´t have a CUDA GPU or not enough Ressources? Now you can run Meshroom on Google Colab. The only thing you need is a Google Account. (+ internet connection with a decent upload rate!)

Note: This solution is a user contribution and not officially supported. For questions and suggestions refer to the original GIST

Google Colab is a free to use Jupyter notebook, that allows you to use free Tesla K80 GPU (24 GB of GDDR5 memory) it also gives you a total of 12 GB of ram, and you can use it up to 12 hours in row(keep the browser tab open as it will reset sooner otherwise). *

---

GPU: 1xTesla K80 , having 2496 CUDA cores, compute 3.7, 12GB(11.439GB Usable) GDDR5 VRAM (some sources state Google now uses Nvidia T4s)

CPU: 1xsingle core hyper threaded i.e(1 core, 2 threads) Xeon Processors @2.3Ghz (No Turbo Boost) , 45MB Cache

RAM: ~12.6 GB Available

Disk: ~320 GB Available

For every 12hrs or so Disk, RAM, VRAM, CPU cache etc data that is on our alloted virtual machine will get erased *

---

Go to https://colab.research.google.com, login and create a new empty python3 notebook. (First steps with Colab)

You NEED to change under Runtime->Change Runtime the selector to GPU. Otherwise it won't work!

Here is a basic sample script by @donmahallem you can use to run Meshroom in the Colab Jupyter notebook:

!git clone https://github.com/alicevision/dataset_buddha

!wget -N https://github.com/alicevision/meshroom/releases/download/v2019.1.0/Meshroom-2019.1.0-linux.tar.gz

!mkdir meshroom

!tar xzf Meshroom-2019.1.0-linux.tar.gz -C ./meshroom

from google.colab import files

# optional upload for the meshfile

uploaded = files.upload()

for fn in uploaded.keys():

print('User uploaded file "{name}" with length {length} bytes'.format( name=fn, length=len(uploaded[fn])))

for fn in uploaded.keys():

print('User uploaded file "{name}" with length {length} bytes'.format( name=fn, length=len(uploaded[fn])))

!mkdir ./buddha_out

# use the --pipeline argument to provide a path to the meshfile you created

!./meshroom/Meshroom-2019.1.0/meshroom_photogrammetry --input ./dataset_buddha/buddha --output ./buddha_out

!tar -czvf meshbuddha.tar.gz ./buddha_out

files.download('meshbuddha.tar.gz')

Link to Gist: https://gist.github.com/donmahallem/22134574382b7bd8a67c1550734fcfc4

If you have written a more sophisticated script, we kindly ask you to share it with the community.

Colab File structure:

Download files:

https://stackoverflow.com/questions/48774285/how-to-download-file-created-in-colaboratory-workspace

https://towardsdatascience.com/downloading-datasets-into-google-drive-via-google-colab-bcb1b30b0166