jmendozais / SDSSDepth

Self-Distillation via Prediction Consistency for Self-Supervised Monocular Depth Estimation

SDSSDepth: Self-Distillation via Prediction Consistency for Self-Supervised Monocular Depth Estimation

SDSSDepth is a codebase that implements the paper Self-Distilled Self-Supervised Depth Estimation in Monocular Videos that will be published at ICPRAI . Also, we hope this codebase could support future research in self-supervised monocular depth estimation and beyond.

Getting Started

Please see Getting Started for guidelines to install and test SDSSDepth.

License

SDSSDepth is released under the MIT License.

Citing SDSSDepth

If you find SDSSDepth useful for your research, please consider citing the following paper:

@inproceedings{mendoza2022self,
	title        = {{Self-Distilled Self-Supervised Monoculer Depth Estimation}},
	author       = {Mendoza, Julio and Pedrini, Helio},
	year         = 2022,
	booktitle    = {International Conference on Pattern Recognition and Artificial Intelligence},
	organization = {Springer}
}

Also, this codebase implements the paper Adaptive Self-Supervised Depth Estimation in Monocular Videos:

@inproceedings{mendoza2021adaptive,
	title        = {{Adaptive Self-supervised Depth Estimation in Monocular Videos}},
	author       = {Mendoza, Julio and Pedrini, Helio},
	year         = 2021,
	booktitle    = {International Conference on Image and Graphics},
	pages        = {687--699},
	organization = {Springer}
}