PlaceFormer: Transformer-Based Visual Place Recognition Using Multi-Scale Patch Selection and Fusion.
Shyam Sundar KannanByung-Cheol MinPublished in: IEEE Robotics Autom. Lett. (2024)
Keyphrases
- multiscale
- image patches
- data fusion
- fuzzy logic
- edge detection
- selection algorithm
- image fusion
- low level
- wavelet transform
- natural images
- visual information
- coarse to fine
- multiple scales
- object recognition
- image features
- selection strategy
- expert systems
- information fusion
- image segmentation
- multi sensor
- fault diagnosis
- fusion method
- deep structure
- visual cues
- neural network
- power system
- d objects
- feature selection