My talk titles are:

Finding needles in a haystack – Road learning Both thin and low frequency objects are difficult for object segmentation, this talk discusses ways to overcome those problems in the road feature learning project. Lidar and cameras are used to capture the data, and projective geometry is used to model and transform the data before applying image learning algorithms.

Using shadows to see – Tomography reconstruction Shadows in 2D can tell us about the object that we are interested in. X-ray and electrons can cast partial shadows of the object, and therefore to image the inside of the 3D object. With suitable data collection, and reconstruction algorithms, we can reconstruct various 3D object represented by cool animations.

We will also connect to Zoom during Thomas's talk if you cannot make it:

https://vuw.zoom.us/my/spgroom