Monocular Depth Distribution Alignment with Low Computation
- URL: http://arxiv.org/abs/2203.04538v1
- Date: Wed, 9 Mar 2022 06:18:26 GMT
- Title: Monocular Depth Distribution Alignment with Low Computation
- Authors: Fei Sheng, Feng Xue, Yicong Chang, Wenteng Liang, Anlong Ming
- Abstract summary: We model the majority of accuracy contrast between light-weight networks and heavy-weight networks.
By perceiving the difference of depth features between every two regions, DANet tends to predict a reasonable scene structure.
Thanks to the alignment of depth distribution shape and scene depth range, DANet sharply alleviates the distribution drift, and achieves a comparable performance with prior heavy-weight methods.
- Score: 15.05244258071472
- License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
- Abstract: The performance of monocular depth estimation generally depends on the amount
of parameters and computational cost. It leads to a large accuracy contrast
between light-weight networks and heavy-weight networks, which limits their
application in the real world. In this paper, we model the majority of accuracy
contrast between them as the difference of depth distribution, which we call
"Distribution drift". To this end, a distribution alignment network (DANet) is
proposed. We firstly design a pyramid scene transformer (PST) module to capture
inter-region interaction in multiple scales. By perceiving the difference of
depth features between every two regions, DANet tends to predict a reasonable
scene structure, which fits the shape of distribution to ground truth. Then, we
propose a local-global optimization (LGO) scheme to realize the supervision of
global range of scene depth. Thanks to the alignment of depth distribution
shape and scene depth range, DANet sharply alleviates the distribution drift,
and achieves a comparable performance with prior heavy-weight methods, but uses
only 1% floating-point operations per second (FLOPs) of them. The experiments
on two datasets, namely the widely used NYUDv2 dataset and the more challenging
iBims-1 dataset, demonstrate the effectiveness of our method. The source code
is available at https://github.com/YiLiM1/DANet.
Related papers
- Progressive Depth Decoupling and Modulating for Flexible Depth Completion [28.693100885012008]
Image-guided depth completion aims at generating a dense depth map from sparse LiDAR data and RGB image.
Recent methods have shown promising performance by reformulating it as a classification problem with two sub-tasks: depth discretization and probability prediction.
We propose a progressive depth decoupling and modulating network, which incrementally decouples the depth range into bins and adaptively generates multi-scale dense depth maps.
arXiv Detail & Related papers (2024-05-15T13:45:33Z) - Diffusion-based Data Augmentation for Object Counting Problems [62.63346162144445]
We develop a pipeline that utilizes a diffusion model to generate extensive training data.
We are the first to generate images conditioned on a location dot map with a diffusion model.
Our proposed counting loss for the diffusion model effectively minimizes the discrepancies between the location dot map and the crowd images generated.
arXiv Detail & Related papers (2024-01-25T07:28:22Z) - Deep Neighbor Layer Aggregation for Lightweight Self-Supervised
Monocular Depth Estimation [1.6775954077761863]
We present a fully convolutional depth estimation network using contextual feature fusion.
Compared to UNet++ and HRNet, we use high-resolution and low-resolution features to reserve information on small targets and fast-moving objects.
Our method reduces the parameters without sacrificing accuracy.
arXiv Detail & Related papers (2023-09-17T13:40:15Z) - Monocular Visual-Inertial Depth Estimation [66.71452943981558]
We present a visual-inertial depth estimation pipeline that integrates monocular depth estimation and visual-inertial odometry.
Our approach performs global scale and shift alignment against sparse metric depth, followed by learning-based dense alignment.
We evaluate on the TartanAir and VOID datasets, observing up to 30% reduction in RMSE with dense scale alignment.
arXiv Detail & Related papers (2023-03-21T18:47:34Z) - Non-parametric Depth Distribution Modelling based Depth Inference for
Multi-view Stereo [43.415242967722804]
Recent cost volume pyramid based deep neural networks have unlocked the potential of efficiently leveraging high-resolution images for depth inference from multi-view stereo.
In general, those approaches assume that the depth of each pixel follows a unimodal distribution.
We propose constructing the cost volume by non-parametric depth distribution modeling to handle pixels with unimodal and multi-modal distributions.
arXiv Detail & Related papers (2022-05-08T05:13:04Z) - Manifold Topology Divergence: a Framework for Comparing Data Manifolds [109.0784952256104]
We develop a framework for comparing data manifold, aimed at the evaluation of deep generative models.
Based on the Cross-Barcode, we introduce the Manifold Topology Divergence score (MTop-Divergence)
We demonstrate that the MTop-Divergence accurately detects various degrees of mode-dropping, intra-mode collapse, mode invention, and image disturbance.
arXiv Detail & Related papers (2021-06-08T00:30:43Z) - Boundary-induced and scene-aggregated network for monocular depth
prediction [20.358133522462513]
We propose the Boundary-induced and Scene-aggregated network (BS-Net) to predict the dense depth of a single RGB image.
Several experimental results on the NYUD v2 dataset and xffthe iBims-1 dataset illustrate the state-of-the-art performance of the proposed approach.
arXiv Detail & Related papers (2021-02-26T01:43:17Z) - Adaptive Context-Aware Multi-Modal Network for Depth Completion [107.15344488719322]
We propose to adopt the graph propagation to capture the observed spatial contexts.
We then apply the attention mechanism on the propagation, which encourages the network to model the contextual information adaptively.
Finally, we introduce the symmetric gated fusion strategy to exploit the extracted multi-modal features effectively.
Our model, named Adaptive Context-Aware Multi-Modal Network (ACMNet), achieves the state-of-the-art performance on two benchmarks.
arXiv Detail & Related papers (2020-08-25T06:00:06Z) - A Single Stream Network for Robust and Real-time RGB-D Salient Object
Detection [89.88222217065858]
We design a single stream network to use the depth map to guide early fusion and middle fusion between RGB and depth.
This model is 55.5% lighter than the current lightest model and runs at a real-time speed of 32 FPS when processing a $384 times 384$ image.
arXiv Detail & Related papers (2020-07-14T04:40:14Z) - Single Image Depth Estimation Trained via Depth from Defocus Cues [105.67073923825842]
Estimating depth from a single RGB image is a fundamental task in computer vision.
In this work, we rely, instead of different views, on depth from focus cues.
We present results that are on par with supervised methods on KITTI and Make3D datasets and outperform unsupervised learning approaches.
arXiv Detail & Related papers (2020-01-14T20:22:54Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.