Related Experiment Video
Updated: Jul 30, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.8K
A Survey on Deep Learning Based Segmentation, Detection and Classification for 3D Point Clouds
Prasoon Kumar Vinodkumar1, Dogus Karabulut1, Egils Avots1
1iCV Lab, Institute of Technology, University of Tartu, 50090 Tartu, Estonia.
Entropy (Basel, Switzerland)
|May 16, 2023
Summary
Deep learning methods are now the top choice for 3D object recognition tasks like segmentation, detection, and classification due to their success in 2D vision. This review assesses the newest deep learning approaches for 3D recognition.
Area of Science:
- Computer Vision
- Machine Learning
- 3D Graphics
Background:
- 3D object recognition is crucial in computer vision, graphics, and machine learning.
- Deep learning has become dominant for 3D segmentation, inspired by 2D computer vision successes.
Purpose of the Study:
- To provide a comprehensive review of recent advancements in deep learning for 3D object recognition.
- To analyze and compare prominent deep learning models used in 3D recognition tasks.
Main Methods:
- Literature review of state-of-the-art deep learning techniques for 3D object recognition.
- Analysis of model performance on benchmark datasets.
- Evaluation of distinctive features of various 3D recognition models.
Main Results:
- Deep learning approaches demonstrate superior performance in 3D segmentation, detection, and classification.
- Numerous innovative deep learning models have been developed and validated.
- Key models exhibit diverse strengths and characteristics.
Conclusions:
- Deep learning is the leading methodology for 3D object recognition challenges.
- The field is rapidly evolving with continuous innovation in model development.
- Understanding model-specific qualities is essential for effective application.
