Lecture 18 - Extending Single View and YOLOΒΆ

Overview:

This lecture builds on the previous lecture and moves to multi-view geometry. Without April tags attached to the back of another vehicle, we take advantage of multi-view geometry from stereo cameras in order to estimate the pose. The second half of this lecture deals with object detection. We will go over YOLO (you only look once), which is a CNN (convolutional neural network) used to classify objects.

Topics Covered:
  • Multi-view geometry

  • YOLO

  • Object detection

Associated Assignment:

Slides:

Links to additional resources: