https://github.com/xiaowuc2/soil-moisture-prediction-on-images-usingtransfer-learning

With transfer learning approach on imbalanced dataset we've achieved 81% accuracy @Inceptionv3
https://github.com/xiaowuc2/soil-moisture-prediction-on-images-usingtransfer-learning

inceptionv3 machine-learning soil-moisture transfer-learning

Last synced: about 1 month ago
JSON representation

With transfer learning approach on imbalanced dataset we've achieved 81% accuracy @Inceptionv3

• Host: GitHub
• URL: https://github.com/xiaowuc2/soil-moisture-prediction-on-images-usingtransfer-learning
• Owner: xiaowuc2
• Created: 2021-11-28T13:38:47.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2021-11-28T13:54:50.000Z (about 3 years ago)
• Last Synced: 2024-01-27T03:01:12.192Z (11 months ago)
• Topics: inceptionv3, machine-learning, soil-moisture, transfer-learning
• Language: Jupyter Notebook
• Homepage:
• Size: 10.2 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        


  

    

  




 Soil moisture prediction on images using Transfer Learning / ​Code​ / ​Website​


### Abstract

Ample moisture levels are of high importance to yields, thus,plants will not grow and develop with inadequate soil moisture which influences soil temperature and heat capacity and also prevents soil fromweathering. Accurate soil moisture detection can lead to proper growthof  all  kind  of  trees  and  crops.  There  are  many  well  established  sensors  which  can  examine  soil  moisture  but  it  is  practically  impossibleto  check  each  field  individually  even  with  sensors  in  a  wide  range  ofarea. In this paper we have used machine learning techniques such as recurrent neural networks, multiple linear regression, VGGNet (VGG-16,VGG-19), Inception v3 etc for prediction of soil moisture for few daysahead  on  field  pictures.  These  techniques  were  applied  on  our  custommade dataset((`4000*1844`) after data augmentation(`200*200`)). Finally,our model classifies images into three different classes‘`Dry`’,‘`Wet`’,‘`Extremely Wet`’