Course Information

Instructor: Prof. Mani Golparvar
Office hours: Tue 2:00pm - 3:00pm or by appointment
Classroom: Newmark 2312 | Time: Tue Thu 3:30pm-5:00pm
Teaching Assistant: Juan Nunez-Morales | Office Hours: Tue Thu 12:30-1:30

Course Overview

This course is an introduction to 2D and 3D visual sensing for data acquisition and analysis of buildings and civil infrastructure systems. It is intended mainly for graduate students who want to acquire basic understanding of the theoretical concepts as well as application of computer vision and image processing for sensing buildings, civil infrastructure systems and sustainable construction operations. Some of the topics introduced are:
a) Image-based 3D reconstruction of construction sites and building components
-- Including 2/3D computer vision topics in image formation, camera models, geometry of multiple views and shape reconstruction methods

b) Location and action tracking of construction equipment and personnel
-- Low-level image processing methodologies such as edge detection, and feature detection
-- Mid-level vision techniques such as segmentation, clustering, and filtering; and
-- Basic high-level vision techniques in equipment and personnel location tracking and action recognition

c) Semantic analysis and intelligent decision making of construction performance and/or operation metrics -- Transforming visual data into knowledge about performance metrics during construction or operation of infrastructure systems

By the end of the course, students will have full understanding of the following concepts and will be prepared for further vision-related investigations with engineering and management applications:
1. Basics of image formation and basic processing: digital images and video streams, camera models and camera calibration techniques
2. Fundamental concepts of single-view metrology, multiple view geometry and structure-from-motion and their application for 3D site reconstruction and recognition
3. Basics of image processing, filters, detectors and descriptors
4. Concepts of image and video segmentation/grouping, clustering and filtering
5. Concepts of component, equipment and personnel recognition (feature-based and region-based classifications)
6. Basics of machine learning techniques for interpreting visual features
7. Range, scope and advantages of visual sensing techniques for monitoring construction progress, productivity, safety, quality and carbon footprint of operations in addition to structural health monitoring and stability analysis

The development of this course in its earlier days was sponsered by MathWorks. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the instructor and do not necessarily reflect the views of MathWorks.