SPEGTI: Structured Prediction for Efficient Generative Text-to-Image Models.
Sadeep JayasumanaDaniel GlasnerSrikumar RamalingamAndreas VeitAyan ChakrabartiSanjiv KumarPublished in: CoRR (2023)
Keyphrases
- structured prediction
- discriminative learning
- efficient learning
- input image
- image segmentation
- image classification
- probabilistic model
- segmentation method
- conditional random fields
- image matching
- latent variables
- bayesian framework
- image retrieval
- prior information
- generative model
- belief networks
- information retrieval
- convex optimization
- markov networks
- maximum margin
- similarity measure
- learning algorithm