Deriving Visual Semantics from Spatial Context: An Adaptation of LSA and Word2Vec to generate Object and Scene Embeddings from Images.
Matthias Sebastian TrederJuan Mayor-TorresChristoph TeufelPublished in: CoRR (2020)
Keyphrases
- spatial context
- image regions
- visual appearance
- multiple objects
- observed scene
- multiple images
- spatial relationships
- complex scenes
- input image
- acquired images
- real world scenes
- real world objects
- visual input
- visual data
- real objects
- three dimensional
- visual scene
- image data
- target object
- viewing angle
- image set
- contextual information
- object models
- spatial relations
- spatial information
- visual features
- temporal context
- image features
- combining information from multiple
- image segments
- visual information
- image classification
- video sequences
- lighting conditions
- single image
- image database
- keypoints
- low level
- moving objects
- category specific
- object shape
- object tracking
- particle filter
- similarity measure