Note board ([Name Redacted].github.io)
This site (repository) provides setup commands (with docker) to run the implementation of papers or packages and the paper clippings . If you want to know this contents, Please check my github.io or Example (docker command) section.
🖥️ Setup commands
Setup commands section notes how to build environments to run the implementation of papers or packages with Docker. Setup commands are written in 'How to build with docker and run the model in a docker container' section.
Example
For example, text2room contains the following items:
- Create a docker container
- Setup packages
- Setup the models
- Run the model
Each item has commands for setup, an example of text2room '1. Create a docker container' is as follows:
# Set this repository absolute path (ex: /home/user/[Name Redacted].github.io)
git clone https://github.com/[Name Redacted]/[Name Redacted].github.io.git
cd [Name Redacted].github.io
OGI_DIR_PATH=$PWD
# Get a base image
BASE_IMAGE=nvidia/cuda:11.7.1-cudnn8-devel-ubuntu20.04
docker pull $BASE_IMAGE
# Create and move to a container dir
mkdir containers
cd containers
# Clone the repository
git clone https://github.com/lukasHoel/text2room.git
# Move to text2room
cd text2room
# Switch to 2023/07/16 ver. git switch -d c38d97e4d418cb5a93cfdc7b89ea0e6e7bbcf20b
# Copy a folder for building env.
📝 Paper clippings and note
Paper clippings section notes summary of paper methods. Paper clippings are written in 'どんなもの?'
Example
An example of text2room 'どんなもの?' is as follows:
- The authors present Text2Room, a method for generating textured 3D meshes of room-scale indoor scenes from a given text prompt as input.
-
In very recent concurrent work, Fridman et al. [18] create 3D scenes from text, but focus on this type of 3D-consistent “zoom-out” video generation.
-
Instead, we generate complete, textured 3D room geometry from arbitrary trajectories.
-
-
We visualize this workflow in Figure 2.
Figure 2. Method overview.
Figure 3. Iterative scene generation.
Note: I used 'どんなもの?' 先端技術とメディア表現1 #FTMA15 . Note: I used 'どんな論文か?', '新規性', '結果' with reference to cvpaperchallenge format .