search

facebookresearch / OccupancyAnticipation

This repository contains code for our publication "Occupancy Anticipation for Efficient Exploration and Navigation" in ECCV 2020.
MIT License
76 stars 26 forks source link

Replicating Habitat Challenge results #11

Closed rpartsey closed 3 years ago

rpartsey commented 4 years ago

Hi dear @srama2512 I followed your README.md instructions and were able to run OccupancyAnticipation evaluation using specified configs without any issues.

The only thing I'd like to ask you is to share the Dockerfile and the agent.py you used for evalai evaluation.

I tried to create my own Dockerfile (and agent.py) but all my attempts to set up the python environment and install OccupancyAnticipation dependencies inside the docker failed.

srama2512 commented 4 years ago

Hi @rpartsey

Unfortunately, I haven't had the chance to create the challenge evaluation script for the latest version of the code. The habitat-api, habitat-sim versions are also not consistent with the challenge version. So I may not be able to provide the script immediately (juggling with a few deadlines).

Coding up the agent should be relatively easy. The code is very similar to the navigation training script that exists in the repo. The biggest issues I faced was installing the right versions of different libraries and getting CUDA to work. So please let me know if you have any specific errors that you are facing and I can help you resolve it. I can provide an older version of my Dockerfile over email if that helps.

rpartsey commented 4 years ago

Thanks for the quick reply. The older version of Dockerfile would be of great help to me. My email is partsey2412@gmail.com.

rpartsey commented 3 years ago

Resolved. Closing the issue.