Learning Topology From Synthetic Data for Unsupervised Depth Completion

Wong, Alex; Cicek, Safa; Soatto, Stefano

doi:10.1109/lra.2021.3058072

Cited by 40 publications

(54 citation statements)

References 34 publications

(174 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to our analysis, in the NConv [13] and Sparse-to-Dense [16] models, the details in the upper part of the output dense depth image are noisy and poor. ScaffNet [40] only used the synthetic data for training and evaluation, which might be not effective when evaluating with real scenario data. The results of the CSPN [35] and Sparse-to-Dense [16] models still include some regions that are incompletely filled.…”

Section: B Results Evaluationmentioning

confidence: 99%

Wasserstein Generative Adversarial Network for Depth Completion With Anisotropic Diffusion Depth Enhancement

Nguyen

Yoo

2022

IEEE Access

View full text Add to dashboard Cite

The objective of depth completion is to generate a dense depth map by upsampling a sparse one. However, irregular sparse patterns or the lack of groundtruth data caused by unstructured data make depth completion extremely challenging. Sensor fusion using both RGB and LIDAR sensors can help produce a more reliable context with higher accuracy. Compared with previous approaches, this method takes semantic segmentation images as additional input and develops an unsupervised loss function. Thus, when combined with supervised depth loss, the depth completion problem is considered as semi-supervised learning. We used an adapted Wasserstein Generative Adversarial Network architecture instead of applying the traditional autoencoder approach and post-processing process to preserve valid depth measurements received from the input and further enhance the depth value precision of the results. Our proposed method was evaluated on the KITTI depth completion benchmark, and its performance in depth completion was proven to be competitive.

show abstract

Section: B Results Evaluationmentioning

confidence: 99%

Wasserstein Generative Adversarial Network for Depth Completion With Anisotropic Diffusion Depth Enhancement

Nguyen

Yoo

2022

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Supported by the progression in learning field, the identification of the VSNs topological features continues to represent a stimulating research topic (see, e.g. [18,19,20]). It is worth to notice that the interest is still mainly directed on the study of the coverage overlap models and the corresponding graph-based representations.…”

Section: Related Workmentioning

confidence: 99%

Visual Sensor Network Stimulation Model Identification via Gaussian Mixture Model and Deep Embedded Features

Varotto,

Fabris,

Michieletto

et al. 2022

Preprint

View full text Add to dashboard Cite

Visual sensor networks constitute a fundamental class of distributed sensing systems, with unique complexity and performance research subjects. One of these novel challenges is represented by the identification of the network stimulation model (SM), which emerges when a set of detectable events trigger different subsets of the cameras. In this direction, the formulation of the related SM identification problem is proposed, along with a proper network observations generation method. Consequently, an approach based on deep embedded features and soft clustering is leveraged to solve the presented identification problem. In detail, the Gaussian Mixture Modeling is employed to provide a suitable description for data distribution and an autoencoder is used to reduce undesired effects due to the so-called curse of dimensionality. Hence, it is shown that a SM can be learnt by solving Maximum A-Posteriori estimation on the encoded features belonging to a space with lower dimensionality. Lastly, numerical results are reported to validate the devised estimation algorithm.

show abstract

“…Whereas, [34] utilized a learned prior (a separate network trained on ground-truth depth) to regularize predictions. [30] learned a topology prior on the sparse points from synthetic data and used it as regularization. We note that supervision from a network trained on a specific domain (e.g.…”

Section: Related Workmentioning

confidence: 99%

An Adaptive Framework for Learning Unsupervised Depth Completion

Wong

Fei

Hong

et al. 2021

IEEE Robot. Autom. Lett.

Self Cite

View full text Add to dashboard Cite

We present a method to infer a dense depth map from a color image and associated sparse depth measurements. Our main contribution lies in the design of an annealing process for determining co-visibility (occlusions, disocclusions) and the degree of regularization to impose on the model. We show that regularization and co-visibility are related via the fitness (residual) of model to data and both can be unified into a single framework to improve the learning process. Our method is an adaptive weighting scheme that guides optimization by measuring the residual at each pixel location over each training step for (i) estimating a soft visibility mask and (ii) determining the amount of regularization. We demonstrate the effectiveness our method by applying it to several recent unsupervised depth completion methods and improving their performance on public benchmark datasets, without incurring additional trainable parameters or increase in inference time.

show abstract

Learning Topology From Synthetic Data for Unsupervised Depth Completion

Cited by 40 publications

References 34 publications

Wasserstein Generative Adversarial Network for Depth Completion With Anisotropic Diffusion Depth Enhancement

Wasserstein Generative Adversarial Network for Depth Completion With Anisotropic Diffusion Depth Enhancement

Visual Sensor Network Stimulation Model Identification via Gaussian Mixture Model and Deep Embedded Features

An Adaptive Framework for Learning Unsupervised Depth Completion

Contact Info

Product

Resources

About