In Defense of Pre-trained ImageNet Architectures for Real-time Semantic Segmentation of Road-driving Images.
Marin OrsicIvan KresoPetra BevandicSinisa SegvicPublished in: CoRR (2019)
Keyphrases
- pre trained
- semantic segmentation
- image collections
- input image
- training data
- image database
- superpixels
- image retrieval
- image classification
- object recognition
- control signals
- multiple images
- three dimensional
- pascal voc
- image set
- image features
- training examples
- object categories
- feature points
- illumination conditions
- multiple views
- image annotation
- object classes
- segmentation algorithm
- graph cuts
- d objects
- learning algorithm