Kognic Terminology Guide

In this article we will explain common names used on our app.

Our platform uses a variety of names and terms, which can sometimes be confusing for both newcomers and long-term users. In this article, we want to clarify these terms for you!

What is a Project?

Think of a Project as a well-organized folder you might use at work. This folder helps you keep all your documents and tasks related to a specific goal in one place.

In our app, a Project serves a similar purpose. It groups and organizes all the annotation work that needs to be done. Within this Project, there are detailed instructions called "annotation requests." These requests specify what data needs to be annotated and how it should be done.

So, a Project helps ensure that all tasks are systematically organized, giving you a clear guidance on what needs to be accomplished and how to achieve it.

 


What is a Request?

Imagine you're working on a big art project with your friends. You have a bunch of pictures (that's the Input batch, we will learn more about it later) that you need to color and add stickers to. But before you start, you need to know a few important things:

  1. What needs to be colored: You need to know which pictures to work on. This is like picking out the right coloring book pages.

  2. How to color and decorate: You need instructions on what colors to use and where to put the stickers. This is like someone giving you rules, saying, "Use blue for the sky and put the star stickers on the night pictures." These rules are the Annotation Instruction.

  3. The steps to follow: There might be a specific order you need to do things in, like coloring first, then adding stickers. This is like following a recipe when you're baking cookies. This process is called the Workflow.

  4. Who will do the work: Finally, you need to know who will color which pictures and who will check to make sure everything is done correctly. This is like assigning your friends different pictures to color and then having someone else check that everyone did a good job. This is the Team.

So, an Annotation request is like your art project plan. It tells you what pictures to work on, how to decorate them, the order to do things in, and who will help and check the work. It's a way to make sure everyone knows what to do and how to do it!

 


What's the difference between an Input Batch and an Input?

We understand that for new users, Inputs and Input Batches might seem quite similar. While they are indeed different, they are closely related and work together!

  • Input: An input is a set of data captured by sensors that you want to analyze or annotate. This data could be:

    • A single image, like a photo taken by a camera at a specific moment.
    • A sequence of images, like frames from a video showing a car driving through a city.
    • A video clip, where you're analyzing motion or events over time.
    • A point cloud, which is a 3D representation of the environment created by combining data from multiple sensors, such as a LiDAR sensor that maps out a room or a street in 3D.
  • Input Batch: An input batch is simply a collection of these inputs. Instead of working with just one image or video, you might be dealing with a group of them, all collected together so that they can be processed or annotated as a unit.


What is a Task?

A task is a specific job or action assigned to a team member that tells them what needs to be done with a particular piece of data, like an image or a video.

The type of the task defines the work required, such as adding annotations, correcting mistakes, or reviewing the quality of the annotations.

Here are the types of tasks that might be assigned to you:
  • Annotate: The team member starts with a blank slate and adds the necessary annotations to the data.
  • Correct: This task involves going through existing annotations and making any necessary corrections.
  • QA Correct: Triggered by a quality expert, this task requires further improvements to the annotations.
  • Review Correct: This task is focused on making corrections based on feedback from a reviewer who found issues with the initial annotations.
  • Quality Assure: The team member checks the quality of the annotations, making improvements to ensure they meet the required standards.
  • Review: A quality expert reviews the annotations to determine if they meet the standards, providing feedback or requesting further corrections if necessary.

As the data moves through different stages of processing, it can generate multiple tasks to ensure that the final output is accurate and of high quality. Each task guides the team member on what needs to be done at that specific point in the workflow.

 


What is a Scene?

One of the most simplest concepts to understand, a Scene is essentially the raw visual representation of the inputs you see when you open a task.

A common confusion we see frequently is between a Scene and an Input, so it's worth to note that Inputs and Scenes are not the same!

 

If you  encounter any issues during this process, don't hesitate to contact our support team at support@kognic.com 🏆