Faster R-CNN

Faster R-CNN first finds places where objects might be, then checks those places carefully.

What it does

Faster R-CNN first finds places where objects might be, then checks those places carefully.

Region: A specific area of a photo • Proposal: An AI's guess

How it works

• Scan the image for possible object areas.
• Zoom in on each possible area.
• Decide what object is inside each area.

Uses and applications

• A hospital tool can help point doctors to areas on medical scans.
• A warehouse camera can count boxes on shelves.
• A wildlife camera can find animals in photos.

In short

It is slower than YOLO but can be more careful.

Related methods

• YOLO (You Only Look Once): Detects objects in a single pass, usually faster.
• Mask R-CNN: Adds a pixel mask for each detected object.
• ResNet: A common backbone for extracting image features.

Back to Computer vision · All topics

The interactive demonstration and the Python walkthrough for this topic need JavaScript.