A novel construction method of convolutional neural network model based on data-driven.
Guofeng ZouGuixia FuMingliang GaoJin ShenLiju YinXianye BenPublished in: Multim. Tools Appl. (2019)
Keyphrases
- data driven
- detection method
- objective function
- similarity measure
- prior knowledge
- high accuracy
- synthetic data
- pairwise
- significant improvement
- main contribution
- machine learning
- segmentation method
- experimental study
- error rate
- clustering method
- edge detection
- computational cost
- cost function
- model selection
- computationally efficient
- dynamic programming
- image sequences
- feature selection
- neural network
- matching algorithm
- high precision
- fully automatic
- optimization method
- evaluation method