What is Object Detection?

A Way for Computer Vision Model to Detect Objects in files

What does an Object Detection Model do?

The Model finds objects of interest in file

We have a bounding box, which is one for each object in the file

Each bounding box has confidence scores

Confidence scores is a probability of the object

Detectors:

We have one-stage and two stage object detectors

One stage object detectors does object detection in one processes

YOLO, SSD, SqueezeDet and DetectNet are one stage object detectors

Two stage object detectors does object detection in two processes

Fast R-CNN is an example of two stage object detection

How does it work?

  1. Input is an Image

  2. Output is probability distribution of classes

We can compare classifier model vs object detection model

In the figure below, A Classifier recognises classes with certain amount

Screen Shot 2022-01-14 at 1.29.05 PM.png

The image below is an object detection model

It gives us where objects are located in image

Advantage of Object detection model is that, it can ignore rest of the content, other than the bounding boxes