Intelligent Image Recognition System for Marine Fouling Using Softmax Transfer Learning and Deep Convolutional Neural Networks

Table 4

Classification results of fouling (first and second training).

Classes	First training (%)	Second training (%)	Percent of improvement (%)
Rock oysters	75.945	96.236	20.291
Kelp	83.817	99.490	15.673
Zebra mussel	77.143	98.350	21.207
Hydrozoan	58.841	92.032	33.191
Acorn barnacle	62.001	96.771	34.770
Algae	58.749	89.231	30.482
Finger sponge	39.369	80.767	41.398
Gooseneck barnacle	89.061	99.903	10.842
Christmas tree worm	65.098	95.908	30.810
Tunicates	86.042	96.344	10.302