A new multi-decision based on Bayesian decision tree algorithm for ship recognition

Ship recognition is a key topic for intelligent shipping in canal. A new multi-decision algorithm based on Bayesian decision tree is proposed. The algorithm combine three Convolutional Neural Networks module - VGG, VGG19 and ResNet to get a better classify result. A quantization method is adopt to quantize the output of these network. A new method based on decision tree construct algorithm is proposed to improve the accurate of decision tree. In experiment, implement a various compared algorithms to prove the algorithm has better performance. (paper)