Introduction to Object Detection Using Deep Learning

(1) Course Outline

This course will consist of three sessions of lectures, covering everything from acquiring basic knowledge of object detection to trying out object detection on arbitrary images. The course will be conducted in accordance with the following three goals: 1.

1. to learn the overview of object detection tasks and evaluation metrics at the implementation level

2. to learn representative methods of object detection, tracing their history

3. experiment with object detection and related tasks using object detection libraries

Object detection has a variety of applications, such as human detection for automatic driving or lesion detection for medical imaging. This course is intended for students who have already taken “Introduction to Deep Learning” or have equivalent knowledge in order to ensure smooth understanding of the lectures, but those who wish to try out object detection are also welcome.

(2) Contents of each session

Session 1: “Overview of Object Detection: History, Tasks, and Applications” 11/16

The first session of the lecture will provide an overview of object detection. Specifically, the lecture will explain object detection tasks (input/output) and evaluation metrics for object detection, comparing them with image recognition tasks, with the goal of grasping the overall picture of object detection. In addition, examples of how object detection is used in the real world to solve problems and its applications will be introduced. After a brief review of CNN, the main method of neural networks for images, VGGNet and ResNet, which are the basic structure of CNN, will be explained with their implementations.