Year: 2023
Type: article
Abstract: The demand of agricultural product has been steadily increased by times. This condition requires related industry to increase their productivity. Automation in smart farming is one of the currently discussed solutions to develop a productive and sustainable farming solution. In this research, we introduce a tool that might help the automation process in farming industry by speeding up the ripeness detection while increasing the accuracy. We designed and developed a system that is able to detect fruit’s ripeness, specifically strawberry. It was built based on machine learning using CNN algorithms. We developed the model using Keras library. We performed the training by feeding 200 images of strawberry fruit with various ripeness levels that covers the entire growth cycle of the fruit. Additionally, these images also were taken from various level of distance to simulate harvester point of view during harvesting. Besides, we performed automated preprocessing to the image data by converting it to HSI color domain. We also built and evaluated the most optimal CNN architecture to retrieve the best training result. The trained model is fed into our own desktop-based application. The testing is performed by feeding more data to the application. The test data shows that our system was able to predict correctly 92-99% in HSI color spectrum compared to 61-67% accuracy in RGB format.
Institution: Universitas Sumatera Utara
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: diamond
OpenAlex ID: W4390592016
