Developing a deep neural network model for predicting carrots volume

In this paper, a deep learning approach to predict carrots volume according to the physical properties was designed. A total of 464 carrots were used for volume prediction. The used carrots were taken from Kasinhani, Konya. First, the data was produced. For this, the length, the diameters with 5 cm intervals, and the volume of each carrot were measured and recorded. The measurements were done using a steel ruler, a vernier caliper, and a glass graduated cylinder. Two deep learning methods: DFN and LSTM were developed to predict carrot volume. The developed systems were implemented with the Keras library for Python. Statistical measures such as Root Mean Squared Error, Mean Absolute Error, and R-2 were used to determine the predicting accuracy of the system. Both methods produced very close values. DFN and LSTM networks achieved 0.9765 and 0.9766 R-2, respectively. RMSE values were 0.0312 for both models. The results obtained showed that both DFN and LSTM are successful and applicable to this task.

Automated summary

In this study, a deep learning approach was developed to predict the volume of carrots based on their physical properties, with both DFN and LSTM networks achieving high predicting accuracy. The statistical measures used confirmed the success and applicability of the developed systems.