depth_estimation

Depth Estimation Benchmark

Introduction

Depth Estimation using monocular images is an emerging field of research and hos shown interesting progress.

Datasets

• NYUDepthV2: Nathan Silberman, Pushmeet Kohli, Derek Hoiem, Rob Fergus, Indoor Segmentation and Support Inference from RGBD Images, ECCV 2012, https://cs.nyu.edu/~silberman/datasets/nyu_depth_v2.html

Models

Fast Depth

• Models Link - NYUDepthV2
• Additional information
• Training Code
• Note: Depth estimation from monocular image input

Dataset	Model Name	Input Size	GigaMACs	Delta1%	Available	Notes
NYUDepthV2	Fast Depth	224x224	0.3825	77.1	Y

MiDaS

• Models Link - NYUDepthV2
• Training Code
• Note: Depth estimation from monocular image input

Dataset	Model Name	Input Size	GigaMACs	Delta1%	Available	Notes
NYUDepthV2	MiDaS-small (v2.1)	256x256	4.633	86.67	Y

References

[1] NYUDepthV2: Nathan Silberman, Pushmeet Kohli, Derek Hoiem, Rob Fergus, Indoor Segmentation and Support Inference from RGBD Images, ECCV 2012, https://cs.nyu.edu/~silberman/datasets/nyu_depth_v2.html

[2] Wofk, Diana and Ma, Fangchang and Yang, Tien-Ju and Karaman, Sertac and Sze, Vivienne, FastDepth: Fast Monocular Depth Estimation on Embedded Systems, IEEE International Conference on Robotics and Automation (ICRA), 2019

[3] Rene Ranftl and Katrin Lasinger and David Hafner and Konrad Schindler and Vladlen Koltun, Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-shot Cross-dataset Transfer, IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2020