Computer vision is one of the most quickly advancing domains of machine learning and perhaps one of the most successful applications of deep learning techniques and convolutional architectures. Image and video data has traditionally been unanalyzable accept for only the most lightweight at specific classifications primarily because of the size of data and its dimensionality. Distributed deep learning techniques have unlocked the information contained in this data precisely because they are designed to handle data of this scope. Applications such as automatic captioning, image search, robotics, and monitoring all show the power of deep learning on images.
In this course, we will look at two deep learning techniques on image data: image classification and object detection. Using transfer learning and ImageNet, students will see how to quickly develop a multi-class image classifier on a custom data set. We will then extend the course to explore using MobileNet to recognize different objects inside a scene. Finally, we will discuss and experiment with facial recognition techniques and consider the implications of image classification and object detection in a real world context.
Upon successful completion of the course, students will be able to:
Understand the impact on privacy and security of scalable image analysis and machine vision.
Demonstrate the use of deep neural networks and transfer learning in image classification using ImageNet.
Explore use of MobileNet to implement object detection in still images and videos.
Experiment with Amazon Rekognition for facial recognition.
Enrollment in this course is open to all students and applies CEUs toward the Machine Learning or Ethics track.
Applies Towards the Following Certificates
- Certificate in Advanced Data Science : Elective Course