Managing labeling projects
1. What is Managing labeling projects?
Labeling projects are spaces where you can organize and manage data for labeling specific datasets. Projects allow you to perform a variety of functions, including creating projects, creating and managing label classes, verifying labeling status, and sharing projects with other workers. Labeling AI's labeling tools and auto-labeling capabilities make data labeling simple.
2. Manage Projects
Dashboards give you a quick, at-a-glance look at the overall overview of your project and manage your own tasks. Create project titles and descriptions, add label classes and name settings, add files, and more. You can also check the progress of the manual labeled image.
1. You can check the progress of the labeling project.
2. Project details
Create and manage project overviews.
You can create project names and descriptions, and automatically update project creation and most recent work days.
Check Manual labeling progress for the entire image file.
You can check how many images are included in each stage of start-up, ongoing, review, and completion.
|Before start||Number of images in progress labelling|
|In Progress||Number of images uploaded but not labeled|
|Review||Number of images examining progressive labeling|
|Done||Number of images that have been labeled at least 1|
|Total||Total Images Uploaded|
Check and manage progress by labeling class. You can create, add, modify, and delete labeling classes. 10 manual labeling is required for each class to proceed with auto labeling. This makes it easy to figure out how much manual labeling has been done based on 10 criteria for each class.
5. Notification History
You can check the artificial intelligence generation process through the notification history.
6. Label Trainers
You can check the number of labeling performed by the labelling worker and each worker.
2-2. Data List
Check the list of images you are working on You can search and catalog images that meet the criteria you set among the uploaded images.
- Select workers: Labeling tasks can be done with members whose projects are shared (to be developed in the future). Select a worker if you want to determine which image has been labeled by each worker.
- Select a job status, view images by job status, and select a job status if you want to collectively delete images for a particular job step.
You can find the file you uploaded with the file name.
You can check the result and modify the label by clicking the labeled data. Please refer to the inspection guide for each model for label modification!
You can add, modify, or delete a result class.
You can select teams and members to share the project with and view the project with your team member's account. Please check the account configuration for additional members.
Set up detailed members to share by setting up roles with groups you added to your account configuration.
The export function of [LABELING AI|Labeling] makes it more convenient to use the data set that has been labeled within DS2.AI and outside. Export supports the ability to develop artificial intelligence immediately with data sets that have been labeled or to store that data on a local computer.
1. Develop artificial intelligence
Artificial intelligence can be developed with labeled datasets in conjunction with [CLICK AI | Modeling] in DS2.AISTUDIO. Artificial intelligence such as prediction, classification, and object recognition can be developed according to the data format. In addition, it can be linked to custom learning that develops artificial intelligence with coding in a jumper environment by renting a desired learning server, and AutoML that automatically develops artificial intelligence without coding.
Download data and label information to your local computer for convenient access outside DS2.AI.
Depending on the type of data, there are storage, COCO storage, and VOC storage.
- Save : Download as a csv file for structured data, natural language data, and for single image classification, download a zip file that stores the image in a folder by class.
Object-type labeling datasets support COCO and VOC formats. Check 'Include Image' on the right side of the Save button to download it with the bounding box and polygon labeled image you worked on.
- COCO : Download labeling information to a JSON file.
- VOC : Download labeling information to an XML file.
You can modify the project name, content, category, etc., and check the notification details.