Code Monkey home page Code Monkey logo

p3i's Introduction

P3I - Perspective Plane Program Induction

This repository contains the official implementation of the paper:

Perspective Plane Program Induction from a Single Image
Yikai Li*, Jiayuan Mao*, Xiuming Zhang, William T. Freeman, Joshua B. Tenenbaum, and Jiajun Wu
In Computer Vision and Pattern Recognition (CVPR) 2020
[Paper] [Project Page] [BibTex]

@inproceedings{Li2020Perspective,
	title={{Perspective Plane Program Induction from a Single Image}},
	author={Li, Yikai and Mao, Jiayuan and Zhang, Xiuming and Freeman, William T. and Tenenbaum, Joshua B. and Wu, Jiajun},
	booktitle={Conference on Computer Vision and Pattern Recognition},
	year={2020}
}

Introduction

P3I is a conceptually simple yet effective algorithm for inducing neuro-symbolic, program-like representation from a single image.

Installation

For the ease of reproducibility, you are suggested to install miniconda (or anaconda if you prefer) before executing the following commands.

git clone https://github.com/42x00/p3i
cd p3i
conda create -y -n p3i
source activate p3i
# Replace cudatoolkit=10.1 with your CUDA version: https://pytorch.org/
conda install -y pytorch cudatoolkit=10.1 -c pytorch
conda install -y pillow opencv

QuickStart

To quickly test P3I on a given image, you can execute

python demo.py --input assets/demo.png

The induced program and rectified image will be saved to the results folder.

Pre-trained Models

You can download our reference pre-trained models from Google Drive. This model was trained on ImageNet by Krizhevsky et al. 2012. We use it to extract visual features from the given image to make the inference procedure more robust.

Induce Program for Your Own Images

To perform P3I on your own images, you need to execute

python demo.py --device 0 --model_ckpt <path-to-pretrained-model> --input <path-to-image> --output_dir <path-to-output>

Here, --device 0 is specifying the GPU ID used for induction, and you can skip it to use CPU.

p3i's People

Contributors

42x00 avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar

Forkers

alwc

p3i's Issues

Use os.mkdirs() to generate folders

In the function visualize_and_dump in p3i\utils\vis.py instead of generating the output folder with:
os.system(f'mkdir -p {args.output_dir}')

try to use the available dedicated function os.mkdirs:
os.makedirs(args.output_dir, exist_ok=True)

On my Windows 10 machine the former command had the side effect of generating a folder named '-p', as I think on Windows the command mkdir does not accept any parameter (https://docs.microsoft.com/en-us/windows-server/administration/windows-commands/mkdir).

Default value for argument 'output_dir'

In the class P3IOptions, the default value for the argument output_dir is currently:
./results
but should be:
results

With the former, the commands os.path.join(args.output_dir, ...) in the function visualize_and_dump (in p3i\utils\vis.py) break on Windows 10, as ./results is a Unix path not valid on Windows.

image

where is the gradient descend

I have went through all the code and still not found some code like loss.backward(). So how does p3i do the gradient descend?

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.