A fully convolutional model towards monocular depth estimation. This project builds heavily from "Unsupervised Monocular Depth Estimation with Left-Right Consistency (Godard, Adoha, Brostow, 2016)" and extends their work by concatenating the output of a semantic segmentation model at the second-last output scale. This produced more visually appealing disparity maps.
My computer vision final project, completed with a partner in my class.