Auto-Rectify Network for Unsupervised Indoor Depth Estimation
This work addresses the challenge of accurate depth estimation in indoor environments for applications like robotics and AR/VR, though it is incremental as it builds on existing unsupervised methods by specifically targeting rotation issues.
The paper tackled the problem of unsupervised depth estimation in indoor videos from handheld devices, where complex ego-motions hinder learning, and proposed an Auto-Rectify Network that automatically rectifies images to remove rotation noise, resulting in performance that significantly outperforms previous state-of-the-art methods on the NYUv2 dataset.
Single-View depth estimation using the CNNs trained from unlabelled videos has shown significant promise. However, excellent results have mostly been obtained in street-scene driving scenarios, and such methods often fail in other settings, particularly indoor videos taken by handheld devices. In this work, we establish that the complex ego-motions exhibited in handheld settings are a critical obstacle for learning depth. Our fundamental analysis suggests that the rotation behaves as noise during training, as opposed to the translation (baseline) which provides supervision signals. To address the challenge, we propose a data pre-processing method that rectifies training images by removing their relative rotations for effective learning. The significantly improved performance validates our motivation. Towards end-to-end learning without requiring pre-processing, we propose an Auto-Rectify Network with novel loss functions, which can automatically learn to rectify images during training. Consequently, our results outperform the previous unsupervised SOTA method by a large margin on the challenging NYUv2 dataset. We also demonstrate the generalization of our trained model in ScanNet and Make3D, and the universality of our proposed learning method on 7-Scenes and KITTI datasets.