3D Reconstruction and Estimation from Single-view 2D Image by Deep Learning – A Survey
Authors:
- 25 June 2024
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Abstract
3D Object Reconstruction from a single-view 2D image has become a promising research field. However, it remains a crucial and unsolved core issue in AI and Computer Vision research. Many scholars think it is the future of Artificial Intelligence and is well deserved at the irreplaceable heart of future AI research. In this paper, the research history of 3D object reconstruction is introduced, and the current state-of-the-art research methods and most novel results are investigated and discussed. A prediction for the best research methods and reconstruction model for this field is made. This paper also provided the essential clues and trends in 2D images or scenes inverse to 3D sceneries by 3D reconstruction. This interdisciplinary research area requires the researchers to have rich knowledge in but not limited to Computer graphics (Such as OpenGL), deep learning, Computer vision (like OpenCV), and the neurocognitive logics and principles of the cerebral cortex.Keywords
This publication has 27 references indexed in Scilit:
- Single-View 3D reconstruction: A Survey of deep learning methodsComputers & Graphics, 2021
- Location Field Descriptors: Single Image 3D Model Retrieval in the Wild2019
- Convolutional Neural Network2018
- Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling2018
- Domain Adaptive Faster R-CNN for Object Detection in the Wild2018
- ScanNet: Richly-Annotated 3D Reconstructions of Indoor Scenes2017
- Deep Residual Learning for Image Recognition2016
- You Only Look Once: Unified, Real-Time Object Detection2016
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Indoor Segmentation and Support Inference from RGBD Images2012