l-dyno: framework to learn consistent visual features using robot's motion.
Kartikeya SinghCharuvahan AdhivarahanKarthik DantuPublished in: CoRR (2023)
Keyphrases
- visual features
- visual information
- visual data
- humanoid robot
- image classification
- visual content
- image search
- motion features
- motion analysis
- image retrieval
- image sequences
- low level
- image collections
- visual appearance
- feature selection
- web images
- visual descriptors
- human actions
- image annotation
- low level features
- vision system
- optical flow
- feature space