This repository contains the segmentation user interface from the OpenSurfaces project, extracted as a lightweight tool

Overview

OpenSurfaces Segmentation UI

This repository contains the segmentation user interface from the OpenSurfaces project, extracted as a lightweight tool. A dummy server backend is included to run the demo.

You can also view the demo online.

To run the demo, there are two versions: one with django, and one with no framework. The django version uses a dummy django server and compiles the website live as necessary. The non-django version is a flat html file extracted from the django version.

If you find this tool helpful, please cite our project:

@inproceedings{bell13opensurfaces,
	author = "Sean Bell and Paul Upchurch and Noah Snavely and Kavita Bala",
	title = "OpenSurfaces: A Richly Annotated Catalog of Surface Appearance",
	booktitle = "SIGGRAPH Conf. Proc.",
	volume = "32",
	number = "4",
	year = "2013",
}

and report any bugs using the GitHub issue tracker. Also, please "star" this project on GitHub; it's nice to see how many people are using our code.

Version 1: Run with Django (Ubuntu Linux)

  1. Install dependencies (coffee-script, django, django-compressor, ua-parser, BeautifulSoup):

    Note: this will change your django current installation if you are not somewhere between 1.4.* and 1.6.*. I suggest looking into the virtualenv package if this is a problem for you.

./django-setup-demo.sh
  1. Start the local webserver:
./django-run-demo.sh
  1. Visit localhost:8000 in a web browser

To get the demo to work on Mac and Windows, you will have to look at the above scripts and run the equivalent commands for your system.

After drawing 6 polygons, the submit button will show you the POST data that would have been sent to the server.

Version 2: Run without Django (Linux or Mac)

  1. Install npm and node.js. On Ubuntu, this is:
sudo apt-get install npm nodejs
  1. Install coffee-script:
sudo npm install -g coffee-script
  1. Build static files (js, css, img) and then start a local python-based webserver:
./python-run-demo.sh
  1. Visit localhost:8000 in a web browser

To get the demo to work on Windows, you will have to look at the above scripts and run the equivalent commands for your system.

Project Notes

POST data

When a user submits, the client will POST the data to the same URL. On success, the client expects the JSON response {"message": "success", "result": "success"}. The client will then notify the MTurk server that the task is completed. For more details, see example_project/segmentation/views.py.

When a user submits, the POST will contain these fields:

results: a dictionary mapping from the photo ID (which is just "1" in
	this example) to a list of polygons.  Example:
	{"1": [[x1,y1,x2,y2,x3,y3,...], [x1,y1,x2,y2,...]]}.
	Coordinates are scaled with respect to the source photo dimensions, so both
	x and y are in the range 0 to 1.

time_ms: amount of time the user spent (whether or not they were active)

time_active_ms: amount of time that the user was active in the current window

action_log: a JSON-encoded log of user actions

screen_width: user screen width

screen_height: user screen height

version: always "1.0"

feedback: omitted if there is no feedback; JSON encoded dictionary of the form:
{
	'thoughts': user's response to "What did you think of this task?",
	'understand': user's response to "What parts didn't you understand?",
	'other': user's response to "Any other feedback, improvements, or suggestions?"
}

Feedback survey

When the user finishes the task, a popup will ask for feedback. In the django version, disable this by setting ask_for_feedback to 'false' in the file example_project/segmentation/vies.py. In the non-django verfsion, update the window.ask_for_feedback variable in index.html.

I recommend asking for feedback after the 2nd or 3rd time a user has submitted, not the first time, and then not asking again (otherwise it gets annoying). Users usually don't have feedback until they have been working for a little while.

Compiling from coffeescript

The javascript for the tool is automatically compiled from coffeescript files by django-compressor and accessed by the client at a url of the form /static/cache/js/*.js. This is set up already if using django.

If not using django, the python-run-demo.sh does this for you by manually compiling coffeescript files and storing them in the /static/ folder.

Browser compatibility

This UI works in Chrome and Firefox only. The Django version includes a browser check that shows an error page if the user is not on Chrome or Firefox or is on a mobile device.

Local /static/ folder

After you run the demo setup, the directory /static/ will contain compiled css and javascript files.

If you are usikng django and change any part of the static files (js, css, images, coffeescript), you will need to repopulate the static folder with this command:

example_project/manage.py collectstatic --noinput

If you are building on top of this repository:

In example_project/settings.py:

  1. Change SECRET_KEY to some random string.
  2. Fill in the rest of the values (admin name, database, etc).

If you want to add this demo to your own (separate) Django project:

In your settings.py file, make the following changes:

  1. Make sure STATIC_ROOT is set to an absolute writable path.

  2. Add this to the STATICFILES_FINDERS tuple:

	'compressor.finders.CompressorFinder',
  1. Add this to the INSTALLED_APPS tuple:
	'django.contrib.humanize',
	'compressor',
	'segmentation',
  1. Add this to settings.py (e.g. at the end):
	# Django Compressor
	COMPRESS_ENABLED = True
	COMPRESS_OUTPUT_DIR = 'cache'
	COMPRESS_PRECOMPILERS = (
		('text/coffeescript', 'coffee --bare --compile --stdio'),
		('text/less', 'lessc -x {infile} {outfile}'),
	)
Owner
Sean Bell
CEO and Co-Founder, GrokStyle Inc. PhD, Cornell University
Sean Bell
Implements Stacked-RNN in numpy and torch with manual forward and backward functions

Recurrent Neural Networks Implements simple recurrent network and a stacked recurrent network in numpy and torch respectively. Both flavours implement

Vishal R 1 Nov 16, 2021
FasterAI: A library to make smaller and faster models with FastAI.

Fasterai fasterai is a library created to make neural network smaller and faster. It essentially relies on common compression techniques for networks

Nathan Hubens 193 Jan 01, 2023
Authors implementation of LieTransformer: Equivariant Self-Attention for Lie Groups

LieTransformer This repository contains the implementation of the LieTransformer used for experiments in the paper LieTransformer: Equivariant self-at

35 Oct 18, 2022
DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective.

DeepSpeed+Megatron trained the world's most powerful language model: MT-530B DeepSpeed is hiring, come join us! DeepSpeed is a deep learning optimizat

Microsoft 8.4k Dec 28, 2022
A simple pytorch pipeline for semantic segmentation.

SegmentationPipeline -- Pytorch A simple pytorch pipeline for semantic segmentation. Requirements : torch=1.9.0 tqdm albumentations=1.0.3 opencv-pyt

petite7 4 Feb 22, 2022
Implementation of Neural Distance Embeddings for Biological Sequences (NeuroSEED) in PyTorch

Neural Distance Embeddings for Biological Sequences Official implementation of Neural Distance Embeddings for Biological Sequences (NeuroSEED) in PyTo

Gabriele Corso 56 Dec 23, 2022
Unofficial implementation of MLP-Mixer: An all-MLP Architecture for Vision

MLP-Mixer: An all-MLP Architecture for Vision This repo contains PyTorch implementation of MLP-Mixer: An all-MLP Architecture for Vision. Usage : impo

Rishikesh (ऋषिकेश) 175 Dec 23, 2022
Differentiable Surface Triangulation

Differentiable Surface Triangulation This is our implementation of the paper Differentiable Surface Triangulation that enables optimization for any pe

61 Dec 07, 2022
The official code for PRIMER: Pyramid-based Masked Sentence Pre-training for Multi-document Summarization

PRIMER The official code for PRIMER: Pyramid-based Masked Sentence Pre-training for Multi-document Summarization. PRIMER is a pre-trained model for mu

AI2 111 Dec 18, 2022
GANsformer: Generative Adversarial Transformers Drew A

GANformer: Generative Adversarial Transformers Drew A. Hudson* & C. Lawrence Zitnick Update: We released the new GANformer2 paper! *I wish to thank Ch

Drew Arad Hudson 1.2k Jan 02, 2023
Semantic Image Synthesis with SPADE

Semantic Image Synthesis with SPADE New implementation available at imaginaire repository We have a reimplementation of the SPADE method that is more

NVIDIA Research Projects 7.3k Jan 07, 2023
Data-Driven Operational Space Control for Adaptive and Robust Robot Manipulation

OSCAR Project Page | Paper This repository contains the codebase used in OSCAR: Data-Driven Operational Space Control for Adaptive and Robust Robot Ma

NVIDIA Research Projects 74 Dec 22, 2022
Multi-Scale Geometric Consistency Guided Multi-View Stereo

ACMM [News] The code for ACMH is released!!! [News] The code for ACMP is released!!! About ACMM is a multi-scale geometric consistency guided multi-vi

Qingshan Xu 118 Jan 04, 2023
DexterRedTool - Dexter's Red Team Tool that creates cronjob/task scheduler to consistently creates users

DexterRedTool Author: Dexter Delandro CSEC 473 - Spring 2022 This tool persisten

2 Feb 16, 2022
Sharpened cosine similarity torch - A Sharpened Cosine Similarity layer for PyTorch

Sharpened Cosine Similarity A layer implementation for PyTorch Install At your c

Brandon Rohrer 203 Nov 30, 2022
PyElastica is the Python implementation of Elastica, an open-source software for the simulation of assemblies of slender, one-dimensional structures using Cosserat Rod theory.

PyElastica PyElastica is the python implementation of Elastica: an open-source project for simulating assemblies of slender, one-dimensional structure

Gazzola Lab 105 Jan 09, 2023
Lazy, a tool for running things in idle time

Lazy, a tool for running things in idle time Mostly used to stop CUDA ML model training from making my desktop unusable. Simply monitors keyboard/mous

N Shepperd 46 Nov 06, 2022
Generalized Decision Transformer for Offline Hindsight Information Matching

Generalized Decision Transformer for Offline Hindsight Information Matching [arxiv] If you use this codebase for your research, please cite the paper:

Hiroki Furuta 35 Dec 12, 2022
RM Operation can equivalently convert ResNet to VGG, which is better for pruning; and can help RepVGG perform better when the depth is large.

RMNet: Equivalently Removing Residual Connection from Networks This repository is the official implementation of "RMNet: Equivalently Removing Residua

184 Jan 04, 2023
TransZero++: Cross Attribute-guided Transformer for Zero-Shot Learning

TransZero++ This repository contains the testing code for the paper "TransZero++: Cross Attribute-guided Transformer for Zero-Shot Learning" submitted

Shiming Chen 6 Aug 16, 2022