Evaluation of Deep Network-based Methods for Crack Detection of Iron Ore Green Pellet

Crack detection for iron ore green pellet is an essential step in the measuring process of drop strength, which one important quality metrics pellet. However, current method crack manual inspection, rather laborious, tedious and subjective. Although various deep network-based methods are proposed to automatically detect cracks tunnel, pavement wall, little effort has been made on detection. Therefore, it still unknown whether can solve problem. In present work, we perform comparison study evaluate performance six state-of-the-art networks, using our dataset with types complex background. Comprehensive comparatives conducted computing efficiency networks Moreover, task-driving performed show what extent affect accuracy strength. Our experimental analyses demonstrate that CrackSegNet achieves better than other five (DeepCrack-Z, DeepCrack-L, U-net, CrackSegNet, GCUnet), thereby performs task strength measurement. time needed by (0.26 seconds per image) longer (0.05–0.20 processing image size 512×512. future needs be improved as well ensure more accurate fast measurement quality.