SceneNet RGB-D: Can 5M Synthetic Images Beat Generic ImageNet Pre-training on Indoor Segmentation?
John McCormacAnkur HandaStefan LeuteneggerAndrew J. DavisonPublished in: ICCV (2017)
Keyphrases
- indoor environments
- image segmentation
- object segmentation
- segmentation method
- linear svm
- segmentation algorithm
- level set
- multiscale
- region growing
- shape prior
- segmentation accuracy
- training process
- fully automatic
- supervised learning
- energy function
- image analysis
- moving shadow
- outdoor scenes
- decision trees
- object categories
- path planning
- domain specific
- training set
- object recognition