Deep Learning (DL) · Intermediate level

Intermediate level · Module 3: Building Image and Text Data Sets · Lesson 1 of 3

Collecting and Labelling Data

Plan what data to collect and write labelling guidelines that produce consistent labels.

In this lesson you will

  • Plan data collection to cover real-world variation
  • Write and apply labelling guidelines

Key concepts

coverage of conditionsclass balancelabelling guidelineslabelling tools

The lesson

Collect data that matches real use: different phones, lighting, angles, backgrounds and partly covered packages. A model only learns the conditions it sees.

Write labelling guidelines with clear class definitions, examples and rules for difficult cases, such as two packages in one photo or a blurry image.

Use a labelling tool or a simple folder structure, and record who labelled each item so problems can be traced.

Step by step

  1. List the conditions the data must cover.
  2. Write the labelling guide with examples.
  3. Collect and label photos in teams.
  4. Record labeller IDs and conditions.

Practice activity

Collect and label the class medicine package data set.

Expected result: A labelled data set with documented coverage and a labelling guide.

Check your understanding

  1. Why collect photos in poor lighting too?

    Show answer

    The model must handle the conditions it will meet in real use.

  2. What should labelling guidelines include?

    Show answer

    Class definitions, examples and rules for difficult cases.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.