What is object detection in deep learning?

Object detection is a deep learning task that identifies what objects are present in an image and where they are located.

Unlike image classification, which assigns a label to the whole image, object detection can find multiple objects and usually draws a bounding box around each one.

What does an object detector predict?

For each detected object, the model typically predicts:

• Class: what the object is.
• Location: where the object appears, usually as bounding-box coordinates.
• Confidence score: how certain the model is about the detection.

For example:

Person → bounding box → 0.94 confidence

Car → bounding box → 0.88 confidence

How does object detection work?

A deep learning model first extracts visual features from the image, often using a CNN or another vision backbone.

It then uses those features to predict object classes and locations.

Modern detectors are commonly grouped into approaches such as:

• One-stage detectors: predict objects directly in a single detection pipeline.
• Two-stage detectors: first generate candidate regions and then classify and refine them.

Examples of object-detection model families include YOLO, SSD, and Faster R-CNN.

Where is object detection used?

Common applications include:

• Autonomous driving
• Surveillance
• Robotics
• Retail analytics
• Medical imaging
• Industrial inspection

Object detection is different from image segmentation, which predicts object regions at the pixel level rather than only placing bounding boxes around them.

For the whole-image task, see “What is image classification in deep learning?”. The pixel-level task is covered separately in “What is image segmentation in deep learning?”.

In short, object detection finds and classifies multiple objects in an image while also predicting where each object is located.