Depth Estimation for Single Image using Hybrid Approach

Abstract

The depth predication is challenging and necessary step, the finding depth relations in single image requires integration of both global and local information. In this work, we propose a method to estimate the depth using U-net architecture. Here we done encoding with resnet-50 and decoding with up-projection architecture. Further, we discussed loss function to estimate the performance. To check the performance the proposed system we used NYU Depth v2 dataset. The proposed system is compared with state of art methods.