An Entire-and-Partial Feature Transfer Learning Approach for Pest Occurrence Frequency

EasyChair Preprint no. 1337

Abstract

The frequency of pest occurrence has always been a task of agricultural time and labor. This paper attempts to solve the above problems through the combination of deep learning and agriculture. We propose an entire-and-partial feature transfer learning architecture to perform pest detection, classification and counting tasks that have the final goal of presenting pest occurrence frequency.
In the partial-feature transfer learning, different fine-grained feature map are strengthened to the entire-feature transfer learning use weight scheme.
 Finally, different fine-grained feature map are strengthened to the entire-feature transfer learning use weight scheme and the cross-layer of the entire-feature network is combined with multi-scale feature map. The entire-feature transfer learning approach enhances the feature map close to the input to the network layer near the output layer, creating a shortcut topology for the input and output layers to reduce the gradient disappearance problem common to deep networks. The experiments result shows that the detection accuracy can be significantly improved and the proposed method can reach 90.2%.

Keyphrases: cross-layer, entire-and partial feature, frequency of pest occurrences, Multi-task Learning., partial-feature transfer learning

@Booklet{EasyChair:1337,
  author = {Chen Yuh-Shyan and Lo Chia-Ling and Hung Hsiang-Ching},
  title = {An Entire-and-Partial Feature Transfer Learning Approach for Pest Occurrence Frequency},
  howpublished = {EasyChair Preprint no. 1337},

  year = {EasyChair, 2019}}