ctwardy / mapscore Goto Github PK
View Code? Open in Web Editor NEWProbability Map Scoring Portal (SARBayes, George Mason, BYU)
Home Page: https://sites.google.com/site/aisearchrescue/
Probability Map Scoring Portal (SARBayes, George Mason, BYU)
Home Page: https://sites.google.com/site/aisearchrescue/
add an anti-reverse engineering page to the new user registration.
Lanny: I thought the textbox size limit for model name is too short (or maybe I just have really long names for my models).
Check with Bob Koester on this.
Lanny: It might also be nice to be able to download the test case scenario with a button. I know I can copy and paste, but I am lazy.
Have the site run some simple models on each case when it's uploaded:
More ambitious (for another task):
implement profiles
Add some sort of captcha to registration.
We're running on the python server now, which is meant for debug/development, not production.
See also the issue "speed".
Make some tests fully available as public tests, so people can see just how the metrics etc work. I guess this would
mean the test case parameters, and both pixel and geo coordinates for IPP and find location.
Add 25%, 50%, 75%, and 95% circles on the test image, just for show.
First version: use "Hiker" or overall average distances.
Second version: use the ones for the correct class.
??Third version: use a Bayes net model??
Lanny: With the “Grid Alignment Verification”, it helps to know what is the size of the grid you are using, because mine might be different. (I just noticed the 5001x5001 in test case. Maybe include this in the instructions for the “Grid Alignment Verification” page?)
Once passwords are secure, have a reminder feature? We can just skype each other for now.
When you have an image already saved in the server, the website often alters this image as part of a url requests and then renders the altered image as part of a template. However, about 30 - 40% of the time, the template will render the previously saved image version without the alterations. you then refresh the page, and the altered image is loaded. It does not seem that a sleep time delay after saving corrects this problem. Thus, we need to find out how to alter, save, and load an image consistently without error.
Bob notes that a uniform gray should get the same score as random, not a -1.
To do this, count all pixels with the same score as the true pixel, and add half above and half below.
Document the change (can be a separate ticket).
Ensure all forms are validated.
(Add "Security" label to the issue tracker here.)
Allow users to delete models.
Add Bob Koester's new cases -- the ones with good coordinates.
Categorize test cases, and in the leaderboard, allow sorting / selecting by category.
Add a quick Leaderboard link to the top menu.
If you happen to find a more pleasing color (white or Mason green perhaps), that's fine too. :-)
Substitute "Scenario" for "Category" on the leaderboard.
Draft an Email for Google regarding % probability rings on satellite images.
Not sure what we meant, but it was in my notes from the 9 July meeting that we should have a beta-test mode.
We shouldn't send the password in cleartext. Find the standard widget/protocol for password handling.
the right hand image formatting on the grid sync pages is off. Repair the html code that dictates how a browser displays the images.
Allow participants to filter cases by country or state -- they may only have data for their own region. Doing this will mean that different groups can challenge different test cases.
Add column sort to leaderboard.
Change "sort" text to "select" or "filter".
Fix typo in "Categroy"
Modify image feedback to use patterns as well as colors.
Use other colors, not red-green, as this is the most common form of color-blindness.
BYU
Paul Dougherty (mobility model)
Naval Postgrad School
Liz @ ESRI
Tom Patterson @ ESRI
Don Ferguson?
MAPSAR group at National Park Service
Consider how to support a scripting interface for batch submission and testing.
use case: Team1 develops a new model and wants it tested on all the cases. Team1 script connects to the site, logs in, registers the new model, and requests to run that model on all test cases. Site does protocol tests and begins sending test cases, one at a time.
When reporting the result, eg .9965, remind people the scale is -1..1, where 0 is random and 1 is perfect.
Site needs text describing the file formats
Site home page should include a short description, then leaderboard, then funding info. Longer description can be below, or via click.
Tried to get the "sarbayes" profile and got a "MultipleObjectsReturned" error. Guess it didn't like both "crt" and "ctwardy".
Environment:
Request Method: GET
Request URL: http://daggre.org:8000/Account_Profile/?Account=SARBayes
Django Version: 1.3
Python Version: 2.6.6
Installed Applications:
['django.contrib.auth',
'django.contrib.contenttypes',
'django.contrib.sessions',
'django.contrib.sites',
'django.contrib.messages',
'django.contrib.staticfiles',
'MapRateWeb.framework',
'django.contrib.admin']
Installed Middleware:
('django.middleware.common.CommonMiddleware',
'django.contrib.sessions.middleware.SessionMiddleware',
'django.middleware.csrf.CsrfViewMiddleware',
'django.contrib.auth.middleware.AuthenticationMiddleware',
'django.contrib.messages.middleware.MessageMiddleware')
Traceback:
File "/home/sarbayes/Python/lib/python2.6/site-packages/Django-1.3-py2.6.egg/django/core/handlers/base.py" in get_response
response = callback(request, _callback_args, *_callback_kwargs)
Active_account = Account.objects.get(institution_name = Account_in)
return self.get_query_set().get(_args, *_kwargs)
% (self.model._meta.object_name, num, kwargs))
Exception Type: MultipleObjectsReturned at /Account_Profile/
Exception Value: get() returned more than one Account -- it returned 2! Lookup parameters were {'institution_name': u'SARBayes'}
Passwords and other possibly sensitive items are sent cleartext in the url!
My profile picture (SARBayes) changed to a giant question mark. Uploading a new image just changed the size of the question mark. But on some screen the old sarbayes logo appears. Are there separate user and organization photos?
We need new test cases, and we need these fields:
The leaderboard should indicate # of test cases for a score (I think it does), alert people to scores based on <10 cases, and possibly estimate the standard deviation or confidence interval for each score.
Suppose #scores = n. One way to estimate is to take other scores based on many cases, create random subsets of size n, and track the spread over those subsets. For N >> n:
Repeat 1000 times:
It may be that 1,000 times is too many, but use it for now.
Use the CSV module to allow reading and writing of .csv files instead of having to delimit them with '|' first.
When user1 is uploading a rating image, the site hangs for user2. It should have better threading, or a pre-emptive "Wait a minute, uploading a rating image for another user..." prompt. Better to provide some sort of concurrency.
add (...)+ to your trailing regexp for the email, or look up standard regexps, to allow c4i.gmu.edu as well as .gmu.edu.
Add a description field for the models. Editable.
Lanny: I see many “Submit Query” buttons. Please rename those accordingly (e.g., Login).
For the Gridsync, implement once per model rather than once per test case.
Create a video tutorial for using the system.
Add "documentation" label to the issue tracker.
Create test images for checking the scores. Uniform, circle in the right area, exact point, etc.
Look into adding NoRobots to site to prevent indexing of database elements.
"Documentation": describe the metrics being used.
Implement
Big task -- implement a unit testing framework and write unit tests for key routines.
Add "Unit testing" label to the issue tracker.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.