Academic Journal
Monocular depth map estimation based on a multi-scale deep architecture and curvilinear saliency feature boosting
| العنوان: | Monocular depth map estimation based on a multi-scale deep architecture and curvilinear saliency feature boosting |
|---|---|
| المؤلفون: | Abdulwahab, Saddam, Rashwan, Hatem A., Masoumian, Armin, Puig, Domènec, Garcia, Miguel Angel |
| المساهمون: | UAM. Departamento de Tecnología Electrónica y de las Comunicaciones |
| بيانات النشر: | Springer |
| سنة النشر: | 2024 |
| المجموعة: | Universidad Autónoma de Madrid (UAM): Biblos-e Archivo |
| مصطلحات موضوعية: | Monocular depth map estimation, Deep autoencoders, Multi-scale networks, Curvilinear saliency, Telecomunicaciones |
| الوصف: | Estimating depth from a monocular camera is a must for many applications, including scene understanding and reconstruction, robot vision, and self-driving cars. However, generating depth maps from single RGB images is still a challenge as object shapes are to be inferred from intensity images strongly affected by viewpoint changes, texture content and light conditions. Therefore, most current solutions produce blurry approximations of low-resolution depth maps. We propose a novel depth map estimation technique based on an autoencoder network. This network is endowed with a multi-scale architecture and a multi-level depth estimator that preserve high-level information extracted from coarse feature maps as well as detailed local information present in fine feature maps. Curvilinear saliency, which is related to curvature estimation, is exploited as a loss function to boost the depth accuracy at object boundaries and raise the performance of the estimated high-resolution depth maps. We evaluate our model on the public NYU Depth v2 and Make3D datasets. The proposed model yields superior performance on both datasets compared to the state-of-the-art, achieving an accuracy of 86% and showing exceptional performance at the preservation of object boundaries and small 3D structures. The code of the proposed model is publicly available at https://github.com/SaddamAbdulrhman/MDACSFB. ; Financial support was given by the pre-doctoral grant (FI 2020) funded by the Catalan government |
| نوع الوثيقة: | article in journal/newspaper |
| وصف الملف: | application/pdf |
| اللغة: | English |
| تدمد: | 0941-0643 |
| Relation: | Neural Computing and Applications; https://doi.org/10.1007/s00521-022-07663-x; Neural Computing and Applications,34.19 (2022): 16423–16440; http://hdl.handle.net/10486/711445; 16423; 19; 16440; 34 |
| DOI: | 10.1007/s00521-022-07663-x |
| الاتاحة: | http://hdl.handle.net/10486/711445 https://doi.org/10.1007/s00521-022-07663-x |
| Rights: | © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2022 ; openAccess |
| رقم الانضمام: | edsbas.8AAF38CB |
| قاعدة البيانات: | BASE |
| تدمد: | 09410643 |
|---|---|
| DOI: | 10.1007/s00521-022-07663-x |