Dual CNN Models for Unsupervised Monocular Depth Estimation

TL;DR

This paper introduces dual CNN models for unsupervised monocular depth estimation, utilizing separate networks for each view and cross disparities, achieving improved results on KITTI and Cityscapes datasets.

Contribution

The paper proposes dual CNN architectures with 6 and 12 loss functions for unsupervised depth estimation, incorporating cross disparities for enhanced accuracy.

Findings

Outperforms recent state-of-the-art on KITTI and Cityscapes datasets.

Dual CNN models effectively utilize cross disparities for better depth estimation.

Code available for reproducibility and further research.

Abstract

The unsupervised depth estimation is the recent trend by utilizing the binocular stereo images to get rid of depth map ground truth. In unsupervised depth computation, the disparity images are generated by training the CNN with an image reconstruction loss. In this paper, a dual CNN based model is presented for unsupervised depth estimation with 6 losses (DNM6) with individual CNN for each view to generate the corresponding disparity map. The proposed dual CNN model is also extended with 12 losses (DNM12) by utilizing the cross disparities. The presented DNM6 and DNM12 models are experimented over KITTI driving and Cityscapes urban database and compared with the recent state-of-the-art result of unsupervised depth estimation. The code is available at: https://github.com/ishmav16/Dual-CNN-Models-for-Unsupervised-Monocular-Depth-Estimation.