Bias-Aware Loss for Training Image and Speech Quality Prediction Models from Multiple Datasets.
Gabriel MittagSaman ZadtootaghajThilo MichaelBabak NaderiSebastian MöllerPublished in: QoMEX (2021)
Keyphrases
- quality prediction
- input image
- multiscale
- low level
- bayesian framework
- image segmentation
- image retrieval
- single image
- image data
- image classification
- image representation
- random fields
- high resolution
- image analysis
- statistical model
- image dataset
- edge detection
- image features
- image quality
- feature points
- region of interest
- acoustic models
- million images
- training and testing data
- image pixels
- machine learning
- image content
- speech recognition
- segmentation algorithm
- probabilistic model
- image collections
- object detectors
- broadcast news
- set of training images
- neural network
- hearing impaired