EURASIP Journal on Advances in Signal Processing

Table 5 Accuracy (%) of seal calls recognition by the CNN model

From: Seal call recognition based on general regression neural network using Mel-frequency cepstrum coefficient features

SNR/dB	MFCC				LOFAR spectrum
SNR/dB	A	B	C	Total	A	B	C	Total
20	98.85	97.13	91.71	94.23	94.51	91.25	90.15	91.97
15	93.61	96.73	90.19	93.51	93.47	90.02	88.37	90.62
10	92.52	95.24	90.61	92.79	89.90	89.88	86.32	88.70
5	90.48	93.46	89.18	91.04	86.75	85.05	80.71	84.17
0	85.91	85.21	85.44	85.52	78.52	75.38	73.89	75.93

Back to article page