mrharicot/monodepth
A Tensorflow implementation of unsupervised single-image depth prediction using convolutional neural networks trained on stereo pairs.
Not currently ranked — collecting fresh signals.
star history
This repository implements a convolutional neural network for monocular depth estimation in an unsupervised manner, using left-right consistency loss as described in the CVPR 2017 paper by Godard et al. The model is trained on stereo image pairs and can predict depth maps from single RGB images. A simplified inference script is provided for running the pretrained model on new images.
Frequently asked
- What is mrharicot/monodepth?
- A Tensorflow implementation of unsupervised single-image depth prediction using convolutional neural networks trained on stereo pairs.
- Is monodepth open source?
- Yes — mrharicot/monodepth is an open-source project tracked on heatdrop.
- What language is monodepth written in?
- mrharicot/monodepth is primarily written in Python.
- How popular is monodepth?
- mrharicot/monodepth has 2.3k stars on GitHub.
- Where can I find monodepth?
- mrharicot/monodepth is on GitHub at https://github.com/mrharicot/monodepth.