Deep Learning (DL) · Advanced level

Advanced level · Module 3: Computer Vision Beyond Classification · Lesson 1 of 4

Object Detection

Find and label several objects in an image with bounding boxes.

In this lesson you will

  • Describe how detectors predict boxes and classes
  • Annotate images and fine-tune a pre-trained detector

Key concepts

bounding boxannotation toolsone-stage and two-stage detectorsnon-maximum suppression

The lesson

Object detection predicts a box and a class for each object in an image. Two-stage detectors propose regions and then classify them; one-stage detectors such as the YOLO family predict boxes directly and are fast.

Detectors output many overlapping boxes; non-maximum suppression keeps the most confident one for each object.

Fine-tuning a pre-trained detector needs annotated boxes. Draw tight, consistent boxes following written guidelines.

Step by step

  1. Annotate 50 trap photos with boxes.
  2. Fine-tune a pre-trained detector.
  3. Visualise predictions with confidence scores.
  4. Count insects per image.

Practice activity

Train the pest detector from the practical activity.

Expected result: A trained detector that draws reasonable boxes and counts on test images.

Check your understanding

  1. What does non-maximum suppression do?

    Show answer

    Removes overlapping boxes, keeping the most confident.

  2. What is the advantage of one-stage detectors?

    Show answer

    They are fast.

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.