New in the Latest 2020 Update
Rapid Annotator 2.0 - 2020 Updated by Gulshan Kumar
During Google Summer of Code 2020, Gulshan Kumar has made improvements and implemented some changes in the Rapid Annotator 2.0. Currently, a test installation can be found here . Please report bugs to Peter Uhrig. The source code can be found on Github. Some indications on how to use the system can be found in Gulshan's GSoC blog.
Latest Features
Notice that in this new release, some new features are available in the homepage.
You can now see the Display Order in the home page as follows:
2. You can now see the your gravatar at the home page.
Admin Panel
Click on the Admin Panel button on the navbar to view the Admin page. (Only Visible to you if you are admin in the Rapid Annotator)
Admin View Page
Here you can see the following information of the Rapid Annotator along with the CRUD of the Admins.
List of Users in the Rapid Annotator
Can Approve/Revoke the Admin Rights Requests.
All the experiments in the Rapid Annotator
Can add a new User in the Rapid Annotator
Able to delete a user from the Rapid Annotator
Can edit the profile of a User
Able to see the profile a User
Progress of particular User
Experiments of a Particular User
All Experiments Page
Here, Admins can see, all experiments present in the rapid annotator. Refer to the below image for more details.
User Profile
Admins can also see each user's profile by clicking on the See Profile button at the admin view page. Refer to the below image for more information.
Experiment Page
Here you will see all the details of the experiment. Click on any experiment at the home page to navigate to the Experiment Page.
1. Add Owner and Add Annotator button has been removed from the experiment view page as it was not semantically aligned and moved to the seetings of the experiment.
2. Pagination (at the bottom of the page) has been added for the displaying the files. Now you can go to last page directly.
Refer to the following images for more details.
Settings Page
Click on the setting button on the experiment to navigate to this page. Here you can do the following things in th new version.
Run the clustering on your data and then data will be displayed according to the clusters
Can selection an option for displaying the target caption or not
Target Caption: These are the user selected caption from the current caption (transcript) as a sub-part using the highligting feature on the annotation page. We will discuss it later in this document.
Can select the no of words per second to display in the caption (transcript) for the unaligned text in the caption.
Can see the experiment progress for all the annotators of the experiment.
Use the following image for more details.
Label View Page
Two features has been added on the label page.
Annotation Level Instruction Field: User can now give instructions about annotating a particular level.
Skip Label Feature: User can select the label as a skip label which means that while annotating if the annotator select that particular label then all the following labels will be skipped. Note: User can select multiple labels as skip labels.
Refer to the following images for details.
View Result Page
The following feature has been added on the view result page.
Pagination
View results for a particular annotator at a time.
Export Results ot .xlsx or .csv format.
For the Concrodances type experiment full concordances will be exported.
Exported results will also take care of annotator's agreement work by displaying the results for each annotator separately.
Selective Annotation: Annotation for a specific file just by clicking on the name of the file at the view result page.
Selective Annotation Process
In the results page you can now directly annotate the specific file whether it was already annotated or not. Follow the below steps for performing selective annotation.
1. First visit the result page. Then click on the filename for the file you want to annotate.
2. If the file is already annotated then it will show the current annotation labels as well as a guidance to the annotator. See the red label button and info on the left.
3. It will also show that if the particular level was skipped or not.
4. After completing the annotation process you will see the following screen. Click on the back button to view the update results
5. Now, you will see the updated results on the result page.
Exported Results
In the exported results, annotation and target caption will be present for each annotator. The word 99999 is RESERVED_LABEL which means that annotation hasn't been done for that file. See the below .xlsx format exported results file as an example.
Annotation Page
The following features has been added to the annotation page.
FullScreen Mode: Now annotators can perform annotation in full screen. Click on Exit FullScreen button for exiting the full screen mode.
Annotation Level Instruction so that annotators can understand what they really need to annotate.
Target Caption
Select the text which you feel as target Caption. It automatically got highlighted then.
Then higlighted text will appear in the text box automatically. if you want you can manually enter the caption in the text area.
The target caption will be saved in the database and you will see them in the exported results.
Caption is aligned according to before and after time of the video.
Check Progress
Now users can check their own progress for each experiment in the following way.
Click on the Check Progress button in your profile section at the navbar.
Now you will see the following window after clicking on the button mentioned in the previous step.
Instructions on Access and Use
Please choose from the three options below how you would like to add your data to the Rapid Annotator. Find out some other new features on the tutorial links:
Manually Uploading Files (Latest), e.g. if you have a set of .mp4 or .jpg files that your would like to upload
Via Spreadsheet (Latest), e.g. if there is a set of files to annotate already available on the web and you have a list of URLs
From a CQPweb Concordance (Latest), if you would like to import a concordance from CQPweb into the Rapid Annotator