DETR — Detection Transformer

Fact

A detection model that uses a Transformer to predict object boxes directly.

In Plain Words

DETR is like a lunch monitor with superhero eyes. One glance at the cafeteria, and it calls out: backpack, banana, mystery meat.

It spots objects and draws boxes in one clean shot. You see ideas like this in self-driving cars and image understanding.

Related Concepts

• Object Detection
• Transformer architecture
• Convolutional Neural Network
• COCO Dataset (Common Objects in Context)

Object Detection
DETR is an end-to-end model for Object Detection.

Transformer
DETR uses a Transformer to see global image relationships.

CNN
Early DETR often used a CNN to extract image features first.

COCO Dataset
COCO Dataset is often used to train and test DETR.