Strategic Improvements of SqueezeSegV2 for Road-Scene Semantic Segmentation Using 3D LiDAR Point Cloud.
Quang-Thai LeQuoc-Hung TranThien Huynh-ThePublished in: SoICT (2023)
Keyphrases
- point cloud
- semantic segmentation
- scene classification
- laser scanner
- stereo camera
- object recognition
- superpixels
- image set
- image classification
- conditional random fields
- d scene
- weakly supervised
- natural scenes
- point sets
- video sequences
- structure from motion
- multiple images
- visual words
- image representation
- object categories
- sparse reconstruction
- multiple objects
- single image
- input image
- image features
- image regions
- object classes
- object class
- high resolution
- viewpoint
- moving objects
- three dimensional
- image processing
- machine learning