Automated and Accurate Counting of Growth Rings in Otoliths Based on Hough Transform and Deep Learning
Souleymane Koné, Abdoulaye Séré, Dekpeltakié Augustin M.S, José Arthur Ouedraogo · 2024
This study presents a significant advance in the field of fish ecology by exploring deep learning to automate and improve the accuracy of counting growth rings in otoliths, which are calcified structures located in the inner ear of fish. Growth rings provide crucial information for estimating the age and growth of fish, but manual counting is prone to human error. We propose an innovative approach based on the use of convolutional neural networks (CNNs) and advanced image processing techniques, particularly the Hough transform.We evaluate the performance of the deep learning model that integrates Hough Transform in the processing step, in order to improve the manual counting of otoliths carried out by experts. The results show that our automated approach achieves comparable or even better accuracy, while considerably reducing analysis time. The use of deep learning and Hough Transform for automated counting of growth rings in otoliths represents a major advance that is revolutionising research into fish ecology and facilitating the effective management of fish population.