VisionAI DataVerse
Observ
English
English
  • Meet DataVerse
  • Data Management
    • Creating Your First Project
    • Import Your Dataset
    • Data Slice - Specific Subsets
    • Data Visualization
      • Image
      • Point Cloud
      • Frame View
      • Sequence View
      • <Use Case> Clean Raw Data
      • <Use Case> Find More Rare Cases
      • <Use Case> Identify Model Weakness
    • Data Metrics
  • Advanced Data Features
    • Image Quality Assessment (IQA)
    • Auto-Tagging
    • Data Discovery (Beta)
    • Data Sampling
    • Data Splitting
    • Data Query
      • Element
      • Logic
      • Use Cases
  • ANNOTATION
    • Before Starting Annotation Task...
    • Create Annotation Task
    • Task Overview
    • Manpower
    • Labeling/Reviewing Panel
      • VQA Labeling Panel
    • Statistics
    • Detail
  • Model Training
    • Train Your AI Model
    • Model Performance
    • Prediction
    • Model Convert (Beta)
    • Model Download (Beta)
  • VisionAI Format
    • VisionAI Data Format
      • coordinate_systems
      • streams
      • contexts
      • objects
      • frames
        • objects
          • bbox
          • cuboid
          • poly2d
          • point
          • binary
        • contexts
        • attributes
      • frame_interval
      • tags
      • metadata
    • Use Case
      • bbox
      • polygon
      • polyline
      • point
      • semantic segmetation
      • classification
      • bbox + cuboid (3d)
      • tagging
    • Format FQA
    • VLM Data Format (VQA)
    • Appendix: Training Format
  • DataVerse Usage
    • Usage and Billing
  • Updates
    • Updates & Release Information
      • Release 2025/4/10
      • Release 2025/1/8
      • Release 2024/11/12
      • Release 2024/09/18
      • Release 2024/08/06
      • Initial Release 2024/01/01
Powered by GitBook
On this page
  • Annotation Task Flow
  • Annotation Task vs. Project
  1. ANNOTATION

Before Starting Annotation Task...

PreviousUse CasesNextCreate Annotation Task

Last updated 10 months ago

Before you start an annotation task, here are some concept that may help you get familiar our annotation module more quickly.

Annotation Task Flow

The current annotation task flow consists one label station and one review station.

A file submitted from the label station will circulate to the review station. If the file is rejected at the review station, it will flow back to the label station. By default, the rejected file will be dispatched to the labeler who submitted it. On the contrary, if the file is approved, it will be marked as done.

As such, a rejected file is defined as in "rework flow" while a file that hasn't or never be rejected is defined as in "initial flow".

Annotation Task vs. Project

The relationship between an annotation task and a project is many-to-one. There can be multiple annotation tasks under a project, and the data slices added to the annotation tasks belong to the corresponding project as well.

Each annotation task allows only one data slice. If you want to label two data slices in one annotation task, combine them on the data slices page.