Abstract:
Many visual tasks use channel attention mechanisms to improve performance. It boosts informative channels and suppresses useless ones. Channel attention modules have been proposed and implemented recently. They rely on convolution and pooling. This paper proposes Gaussian process embedded channel attention (GPCA) module and probabilistic channel attention schemes. GPCA models channel correlations using beta distributed variables. We use a beta distribution approximation to train convolutional neural networks (CNNs) end-to-end because the beta distribution cannot be mathematically integrated. We use a Sigmoid-Gaussian approximation to transfer Gaussian variables into [0,1]. The Gaussian process models channel correlations. This solution is mathematically tractable. CNN end-to-end training can be efficiently implemented with the GPCA module. The GPCA module performs well in experiments.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
