SATBench: Benchmarking the speed-accuracy tradeoff in object recognition by humans and dynamic neural networks.
Ajay SubramanianSara PriceOmkar KumbharElena SizikovaNajib J. MajajDenis G. PelliPublished in: CoRR (2022)
Keyphrases
- neural network
- object recognition
- high accuracy
- execution speed
- processing speed
- computational complexity
- prediction accuracy
- computer vision
- image understanding
- image representation
- genetic algorithm
- error rate
- classification accuracy
- trade off
- d objects
- object detection
- high speed
- computational cost
- pattern recognition
- artificial neural networks
- highly accurate
- recurrent neural networks
- multiscale
- image features
- fuzzy systems
- multi layer
- human subjects
- neural network model
- template matching
- fuzzy rules
- self organizing maps
- mobile robot