Sulfur analysis in apricot with deep learning

The sulfurization process is important for the long-term preservation of dried apricots. However, if the sulfurization process exceeds certain ppm amounts, dried apricots cannot be exported. Therefore, sulfur analysis of apricots is mandatory. Sulfur analysis is usually carried out at apricot research centers, universities and Malatya apricot exchange for a certain fee. This chemical analysis process causes a waste of time for apricot demanders. In the process of sulfur analysis in apricot, a lot of water is consumed and this causes water waste. The necessity to employ chemists requires both qualified manpower and cost. In this thesis, it is aimed to automate this challenging apricot sulfur analysis by using deep learning and image processing techniques, shortening the process, reducing the cost, eliminating water waste and chemical use. In this way, the time loss and labor requirement experienced in the traditional sulfur analysis process are minimized. In addition, since there is no waste of water and chemical use, environmental damage is also eliminated. This automated analysis method provides a significant advantage for apricot exporters and other interested parties. KEYWORDS: Sulfurized apricot, Apricot analysis, Deep learning, Sulfur analysis, AlexNet, DarkNet53, DenseNet201, MobileNetV2, ResNet50