Depth Map Prediction from a Single Image using a Multi-Scale Deep Network

Code

• github.com/SeokjuLee/Insta-DM
  pytorch★ 228
• github.com/MasazI/cnn_depth_tensorflow
  tf★ 127
• github.com/sejong-rcv/2021.Paper.TransDSSL
  pytorch★ 12
• github.com/kieran514/dyna-dm
  pytorch★ 12
• github.com/zhengtr/CS231N-Final-Project
  pytorch★ 0
• github.com/dsshim0125/dacl
  pytorch★ 0
• github.com/imran3180/depth-map-prediction
  pytorch★ 0
• github.com/Tom-Zheng/depth_single_image
  tf★ 0
• github.com/mindspore-ai/models/tree/master/official/cv/depthnet
  mindspore★ 0
• github.com/shuuchen/depth_eigen
  pytorch★ 0

Abstract

Predicting depth is an essential component in understanding the 3D geometry of a scene. While for stereo images local correspondence suffices for estimation, finding depth relations from a single image is less straightforward, requiring integration of both global and local information from various cues. Moreover, the task is inherently ambiguous, with a large source of uncertainty coming from the overall scale. In this paper, we present a new method that addresses this task by employing two deep network stacks: one that makes a coarse global prediction based on the entire image, and another that refines this prediction locally. We also apply a scale-invariant error to help measure depth relations rather than scale. By leveraging the raw datasets as large sources of training data, our method achieves state-of-the-art results on both NYU Depth and KITTI, and matches detailed depth boundaries without the need for superpixelation.

Tasks

Reproductions