Tom Krolikowski

SNACBot

The need for depth perception in the modern world increases dramatically each year. With the creation of self-driving cars and other smart machines, there is an importance in capturing the way human eyes are able to interpret the 3-dimensional world and efficiently convert it into data computers can understand. Using computer vision, and more specifically convolutional neural networks, the problem of mapping real world images into a depth map becomes possible. Using encoder-decoder CNNs, we can capture the spacial relationship between each pixel in an input image from each layer of the net.

Semantic Segmentation Depth Estimation

The need for depth perception in the modern world increases dramatically each year. With the creation of self-driving cars and other smart machines, there is an importance in capturing the way human eyes are able to interpret the 3-dimensional world and efficiently convert it into data computers can understand. Using computer vision, and more specifically convolutional neural networks, the problem of mapping real world images into a depth map becomes possible. Using encoder-decoder CNNs, we can capture the spacial relationship between each pixel in an input image from each layer of the net.