Application of Convolutional Neural Network in Feather Classifications
Milind Shah, Keval Nagda, Anirudh Mukherjee, Pratik Kanani · 2021
This chapter discusses the experimental results indicate that the basic deep models with proper training strategies have more capabilities than what have been explored for multilabel image classification, and a strong baseline. The work also investigates the effectiveness of data augmentation by implementing a different set of operations, and experimental results show superior performance when using data augmentation strategies. A neural network can have many different architectures wherein they tend to work on different sets of problems. A feedforward neural network is one of the most basic kinds of neural networks in which the output of the member neurons is fed forward to the next layer and thereafter the output is evaluated. The work also compares the proposed baseline performance against that of state-of-the-art approaches and without any tricks, only basic deep convolutional neural network models but achieve better performance.