Home

Awesome

DIFFNet

This repo is for Self-Supervised Monocular Depth Estimation with Internal Feature Fusion(arXiv), BMVC2021

A new backbone for self-supervised depth estimation.

PWC

If you think it is a useful work, please consider citing it.

@inproceedings{zhou_diffnet,
    title={Self-Supervised Monocular Depth Estimation with Internal Feature Fusion},
    author={Zhou, Hang and Greenwood, David and Taylor, Sarah},
    booktitle={British Machine Vision Conference (BMVC)},
    year={2021}
    }

Update:

Comparing with others

Evaluation on selected hard cases:

Trained weights on KITTI

Methodsabs relsq relRMSErmse logD1D2D3
1024x3200.0970.7224.3450.1740.9070.9670.984
1024_320_ms0.0940.6784.2500.1720.9110.9680.984
1024x320_ms_ttr0.0790.6403.9340.1590.9320.9710.984
640x1920.1020.7534.4590.1790.8970.9650.983
640x192_ms0.1010.7494.4450.1790.8980.9650.983

Setting up before training and testing

Training:

sh start2train.sh

Testing:

sh disp_evaluation.sh

Infer a single depth map from a RGB:

sh test_sample.sh

Acknowledgement

Thanks the authors for their works: