MAS512 Computer Vision and Embedded AI (Autumn 2024)

On successful completion of the course, the student should be able to:

• describe a camera model and be able to calibrate a camera to determine the camera matrix that describes the mapping from 2D image to the 3D world
• use image processing techniques to filter the image and detect objects in the image
• use sensors such as lidar and radar to make 2D or 3D scans of objects and environments and perform point-cloud processing such as pose estimation
• configure, train, and use artificial intelligence (AI) algorithms for use in vision applications and
• select suitable hardware platform for a given AI task, and deploy a pre-trained neural network to the selected platform.

IMAGE PROCESSING - image filtering, image segmentation, blob detection in images, image representation, transformation between different color spaces, detect and localize landmarks, Fourier transformations, convolution

CAMERA MODELING AND CALIBRATION - Modeling the camera with a camera model, calibrate camera and get the camera matrix and distortion parameters,

3D VISION - SENSORS – LIDAR, RADAR - Lidar, Radar , principles of different sensors for measuring depth, distance, velocity, imaging, point cloud processing, radar signal processing

ALGEBRA AND STOCHASTIC THEORY- Linear Algebra, Probability, Stochastic Theory

MACHINE LEARNING - machine learning architectures (including CNN), configuring and training artificial neural networks, deploying trained artificial neural networks on embedded hardware, RL.

Information about exam requirements will be discussed in canvas during class. See exam section below.