DS-Depth: Dynamic and Static Depth Estimation via a Fusion Cost Volume
نویسندگان
چکیده
Self-supervised monocular depth estimation methods typically rely on the reprojection error to capture geometric relationships between successive frames in static environments. However, this assumption does not hold dynamic objects scenarios, leading errors during view synthesis stage, such as feature mismatch and occlusion, which can significantly reduce accuracy of generated maps. To address problem, we propose a novel cost volume that exploits residual optical flow describe moving objects, improving incorrectly occluded regions volumes used previous work. Nevertheless, inevitably generates extra occlusions noise, thus alleviate by designing fusion module makes compensate for each other. In other words, occlusion from is refined volume, incorrect information eliminated volume. Furthermore, pyramid distillation loss photometric inaccuracy at low resolutions an adaptive direction large gradient regions. We conducted extensive experiments KITTI Cityscapes datasets, results demonstrate our model outperforms previously published baselines self-supervised estimation.
منابع مشابه
Extending Depth of Field via Multifocus Fusion
ii Dedicated to the strength and courage of my mother, and with fond salutations to the cities of Coimbatore, Thamizhnadu (TN) & Knoxville, Tennessee (TN) and the wonderful people that live(d) there and love me. iii Acknowledgements Firstly, before I say anything, I want it heard that I have had a great life here in Tennessee. I am thankful for this period in my life and I am glad I went throug...
متن کاملFusion of Combined Stereo and Focus Series for Depth Estimation
Estimating depth maps is an important step towards 3D reconstruction. In order to compensate the insufficiencies that different image based methods impose, a fusion of multiple information sources for depth estimation is proposed. This article concentrates on obtaining depth information by fusing combined stereo and focus series, which are efficiently gained using a multivariate camera array.
متن کاملImage Guided Cost Aggregation for Hierarchical Depth Map Fusion
Estimating depth from a video sequence is still a challenging task in computer vision with numerous applications. Like other authors we utilize two major concepts developed in this field to achieve that task which are the hierarchical estimation of depth within an image pyramid as well as the fusion of depth maps from different views. We compare the application of various local matching methods...
متن کاملOnline Temporally Consistent Indoor Depth Video Enhancement via Static Structure
• State-F: Forward Outliers p(dx|Zx,mx = 1) = Uf (dx|Zx) = Uf · 1[dtxZx]. For the purpose to combine all the three states into a united model and describe the overall likelihood that the input depth samples fit the current static structure, we use a mixture model similar to the Gaussian Mixture Model [1]. Together with ...
متن کاملImproved Depth Recovery In Consumer Depth Cameras via Disparity Space Fusion within Cross-spectral Stereo
We address the issue of improving depth coverage in consumer depth cameras based on the combined use of cross-spectral stereo and near infrared structured light sensing. Specifically we show that fusion of disparity over these modalities prior to subsequent optimization, within the disparity space image, facilitates the recovery of scene depth information in regions where structured light sensi...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE Transactions on Circuits and Systems for Video Technology
سال: 2023
ISSN: ['1051-8215', '1558-2205']
DOI: https://doi.org/10.1109/tcsvt.2023.3305776