Managing labeling projects tutorial - SentiSight.ai

Managing labeling projects tutorial

Topics covered:
  • Sharing projects
  • Managing roles and permissions
  • Tracking image labeling time
  • Filtering images
  • Reviewing labeled images
  • Labeling settings

You can download video tutorial here


Managing labeling projects tutorial video transcription

  • Sharing projects

    The SentiSight collaboration tools allow you to share projects, so that multiple users can label images at the same time. To share the project, click on the ‘Add User’ button at the top, and then enter the email address of your collaborator.
  • Managing roles and permissions

    Once you have added users to your projects, you can manage their Roles and Permissions, by clicking on the User Settings button, and then on ‘Edit Roles’.

    You can choose between Project Owner, Labeler, Developer or Admin.

    Once you have chosen a role, you can choose from a list of User privileges, including;

    • Adjust validation set
    • Classification Model Training
    • Delete Image
    • Delete Project
    • Download Image labels
    • Download Model
    • Edit Image
    • Edit project settings
    • Edit User Roles
    • Filter Images by user
    • Image Download
    • Image labeling
    • Image Upload
    • Object Detection Model training
    • Prediction
    • Project Sharing
    • See labeling time
    • Segmentation model training
    • Similarity Search
  • Tracking image labeling time

    SentiSight tracks how long each user has spent labeling images. To see the labeling times, click on the Person Icon on the top right corner, and then click on ‘See time’ next to the Labels Made Feature. This information will show the labeling time per user per day, as well as summaries displaying Total Labeling Time per user per project.

    You can filter the data and labeling time by a user basis, project basis, or by the start and end date. You can download these user labeling time statistics as csv files.

  • Filtering images

    SentiSight has a range of ways to filter images to help you with your labelling. You can Filter by Type (left hand menu), and choose from options such as Contains Object Labels, or not.

    You can also filter images by their label name, choosing from either classification labels, object labels, or All.

    Finally, you can choose to “Filter by image status” which allows you to filter images by whether they have been Seen By and Labeled By particular users, to help with project management.

    A new feature on SentiSight is the ‘eye’ icon that appears on the top right corner of the images, that indicates whether the image has been seen by you or not, making it easy to understand your reviewing progress. You can manually choose to set an image as seen or unseen by right clicking on the image and selecting the relevant option. You can select multiple images at once and set their seen or unseen status in the same way.

  • Labeling settings

    There are a range of labeling settings that you can choose from. Simply click on the settings icon in the top right corner, and then choose labeling from the menu.

    You can select the Synchronize labels option, which means that whenever you are adding an object label, this will also appear as a classification label.

    Users are also given the option to Skip already labeled images to help speed up the labeling process. Note that this option is now replaced by “Skip other user images”, because sometimes it is convenient to come back to the previous images that you labeled yourself.

  • Reviewing labeled images

    If you want to skip the images you labeled, too, you can use the “Filter by image status” tool. The “Skip other user images” setting is useful when several people are labeling images in the same project at the same time and, therefore, offline image filtering is not a suitable option.
gdpr-image
We use cookies and other technologies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it..
Privacy policy