North American Hardwoods Identification Using Machine-Learning(SCI 读书笔记)

1. Abstract

   1. how do they write abstracts

      First summarize the Article Technology

      And then briefly describe the specific experimental methods

      including data acquisition , data validation,data set partitioning,data enhancement

      finally,the experimental results are given ,and the future application of the technology is briefly prospected

2. Introduction

   1. How do they write Introduction
      1. Status of wood identification work
      2. the sigificance of the core function of this tehnology
         1. Accurate and rapid wood identification
         2. three paragraphs are used to describe
      3. Used two paragraphs to briefly introduce author's technical background

3. Materials and Methods

   1. Generation of Wood Sample Database
      1. using a paragraph to describe the source of the image
      2. the initial image data set is introduced using a paragraph
   2.  Image Acquisition Setup and Dataset Processing 
      1. using a paragraph to describe the pre-phsical treatment and shooting of wood
      2. another paragraph describes the way image enhancement and validation sets are optimized
   3. Convolutional Neural Network Architecture
      1. describe the specific use of neural networks

4. Results and Discussion

   1. Showing Learning rate decay
   2. Negative argument on overfitting
   3. verification set and training set accuracy image and loss image display
   4. verification set and training set recall data display

5. Conclusions