An image-based deep learning model for water turbidity estimation in laboratory conditions

Feizi H., Sattari M. T., Mosaferi M., Apaydin H.

INTERNATIONAL JOURNAL OF ENVIRONMENTAL SCIENCE AND TECHNOLOGY, cilt.20, sa.1, ss.149-160, 2023 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 20 Sayı: 1
  • Basım Tarihi: 2023
  • Doi Numarası: 10.1007/s13762-022-04531-y
  • Dergi Adı: INTERNATIONAL JOURNAL OF ENVIRONMENTAL SCIENCE AND TECHNOLOGY
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aqualine, Aquatic Science & Fisheries Abstracts (ASFA), Biotechnology Research Abstracts, CAB Abstracts, Compendex, Environment Index, Geobase, INSPEC, Pollution Abstracts, Veterinary Science Database
  • Sayfa Sayıları: ss.149-160
  • Anahtar Kelimeler: Classification, Convolutional neural network, Python programming, Water quality, QUALITY, SYSTEM
  • Ankara Üniversitesi Adresli: Evet

Özet

One of the important parameters in water quality, especially surface water, is the amount of turbidity in the water which can have a harmful effect on irrigation water and drinking water. Existing methods for measuring water turbidity, although relatively simple, are costly due to the need for a turbidity meter. In addition, human intervention or non-calibration of the device can cause errors. Therefore, in this paper, a deep learning method including an image-based convolution neural network is used to estimate water turbidity. In this way, different samples of water turbidity were prepared and photographed in the laboratory and entered the network. The network was implemented using Python programming on the Google Colab platform. The results of the proposed system were compared with the results of the Turbidity-Meter. The results showed that the proposed method's performance is good and at best classifies only one sample in the wrong class. This method can detect the water turbidity class with 97.5% accuracy and only 2.5% error due to lack of complete control of environmental factors in imaging. Consequently, having an extensive database of water quality images in different conditions, it is possible to detect water quality with sufficient accuracy only from images and there is no need to spend a lot of money. This method seems to be able to play an important role in areas where water turbidity is important but lacks laboratory facilities.