Refine and Distill: Exploiting Cycle-Inconsistency and Knowledge Distillation for Unsupervised Monocular Depth Estimation

Pilzer, Andrea; Lathuilière, Stéphane; Sebe, Nicu; Ricci, Elisa

doi:10.1109/cvpr.2019.01000

Cited by 134 publications

(62 citation statements)

References 40 publications

(112 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Specifically, we report in the table competitors for whose the source code or trained models are available, self-supervised either using monocular or stereo images. Unfortunately, code and models are not available for [46,34], and hence, we are not able to compare with them. Methods marked with * have been pre-trained on CityScapes dataset [5], for whose the authors do not provide weights trained on KITTI only.…”

Section: Ablation Studiesmentioning

confidence: 99%

Enhancing Self-Supervised Monocular Depth Estimation with Traditional Visual Odometry

Andraghetti¹,

Myriokefalitakis²,

Dovesi³

et al. 2019

2019 International Conference on 3D Vision (3DV)

View full text Add to dashboard Cite

Estimating depth from a single image represents an attractive alternative to more traditional approaches leveraging multiple cameras. In this field, deep learning yielded outstanding results at the cost of needing large amounts of data labeled with precise depth measurements for training. An issue softened by self-supervised approaches leveraging monocular sequences or stereo pairs in place of expensive ground truth depth annotations. This paper enables to further improve monocular depth estimation by integrating into existing self-supervised networks a geometrical prior. Specifically, we propose a sparsity-invariant autoencoder able to process the output of conventional visual odometry algorithms working in synergy with depth-from-mono networks. Experimental results on the KITTI dataset show that by exploiting the geometrical prior, our proposal: i) outperforms existing approaches in the literature and ii) couples well with both compact and complex depth-from-mono architectures, allowing for its deployment on high-end GPUs as well as on embedded devices (e.g., NVIDIA Jetson TX2).

show abstract

Section: Ablation Studiesmentioning

confidence: 99%

Enhancing Self-Supervised Monocular Depth Estimation with Traditional Visual Odometry

Andraghetti¹,

Myriokefalitakis²,

Dovesi³

et al. 2019

2019 International Conference on 3D Vision (3DV)

View full text Add to dashboard Cite

show abstract

“…The work of [24] integrates heterogeneous predictions from global and local networks. Andrea Pilzer et.al [34] train a student network to predict a disparity map and a backward cycle network for generating image to re-synthesize back the input image. Then the third network exploits the inconsistency between the original and the reconstructed input frame in order to output a refined depth map and knowledge distillation are exploited.…”

Section: B Benchmark Performance 1) Compared Methodsmentioning

confidence: 99%

Soft Regression of Monocular Depth Using Scale-Semantic Exchange Network

Zhang

2020

IEEE Access

View full text Add to dashboard Cite

This paper focuses on depth estimation from single monocular image. Most of existing methods regress depth values or classify depth labels, based on single scale feature representations. However, neither regression nor classification can avoid their inherent defects. Single scale context and low-level semantic cannot support accurate depth estimations. We innovatively address single monocular depth estimation by performing soft regression on probability distribution of classification generated by our proposed scale-semantic exchange network (SSE-Net). Our network maintains rich high-resolution representations. With adding high-to-low resolution to form more stages, the repeated context fusions guarantee each representation receives scale information from other parallel representations over and over. Mutual channel attention mechanism is proposed for emphasizing specific semantics of feature representation. We allow each depth class to be shifted and scaled adaptively. Depth estimation is calculated as expected value of probability distribution. The experimental results verify the efficiency of every proposed components and show competitive results compared with the recent state-of-the-art methods.

show abstract

“…The purpose of knowledge distillation is to transfer the knowledge learned by the teacher network to the student network, so that the functions learned by the large model are compressed into smaller and faster models. Pilzer et al [72] followed this idea, and knowledge distillation is used to transfer information from the refinement network to the student network. Considering the effectiveness of training with synthetic images, Zhao et al [78] adopted the framework of cycle GAN for the transformation between the synthetic and real domains to expand the data set, and propose a geometryaware symmetric domain adaptation network (GASDA) to make better use of the synthetic data.…”

Section: Semi-supervised Monocular Depth Estimationmentioning

confidence: 99%

Monocular depth estimation based on deep learning: An overview

Zhao¹,

Sun²,

Zhang³

et al. 2020

Sci. China Technol. Sci.

203

View full text Add to dashboard Cite

Depth information is important for autonomous systems to perceive environments and estimate their own state. Traditional depth estimation methods, like structure from motion and stereo vision matching, are built on feature correspondences of multiple viewpoints. Meanwhile, the predicted depth maps are sparse. Inferring depth information from a single image (monocular depth estimation) is an ill-posed problem. With the rapid development of deep neural networks, monocular depth estimation based on deep learning has been widely studied recently and achieved promising performance in accuracy. Meanwhile, dense depth maps are estimated from single images by deep neural networks in an end-to-end manner. In order to improve the accuracy of depth estimation, different kinds of network frameworks, loss functions and training strategies are proposed subsequently. Therefore, we survey the current monocular depth estimation methods based on deep learning in this review. Initially, we conclude several widely used datasets and evaluation indicators in deep learning-based depth estimation. Furthermore, we review some representative existing methods according to different training manners: supervised, unsupervised and semi-supervised. Finally, we discuss the challenges and provide some ideas for future researches in monocular depth estimation.

show abstract

Refine and Distill: Exploiting Cycle-Inconsistency and Knowledge Distillation for Unsupervised Monocular Depth Estimation

Cited by 134 publications

References 40 publications

Enhancing Self-Supervised Monocular Depth Estimation with Traditional Visual Odometry

Enhancing Self-Supervised Monocular Depth Estimation with Traditional Visual Odometry

Soft Regression of Monocular Depth Using Scale-Semantic Exchange Network

Monocular depth estimation based on deep learning: An overview

Contact Info

Product

Resources

About