The Impact of Background Removal on Performance of Neural Networks for Fashion Image Classification and Segmentation.
Junhui LiangYing LiuVladimir VlassovPublished in: CoRR (2023)
Keyphrases
- image classification
- neural network
- level set
- image segmentation
- segmentation method
- background pixels
- foreground and background
- multiscale
- region growing
- artificial neural networks
- image representation
- foreground background segmentation
- segmentation algorithm
- complex background
- multi layer
- visual words
- genetic algorithm
- low depth of field
- background modeling
- foreground segmentation
- segmentation accuracy
- shape prior
- fully automatic
- self organizing maps
- medical images
- feature extraction
- medical imaging
- pattern recognition
- object regions
- intensity variations
- global thresholding
- object recognition
- image analysis
- fully unsupervised
- neural network model
- energy function
- class specific
- bag of words
- object segmentation
- background subtraction