Research Article

Underwater Image Processing and Object Detection Based on Deep CNN Method

Table 2

mAP and precision of different iteration times by YOLO v3 and modified methods. ().

IterationOriginal networkScheme 1Scheme 2
mAP (%)Precision (%)mAPPrecision (%)mAPPrecision (%)
Sea cucumberSea urchinScallopSea cucumberSea urchinScallopSea cucumberSea urchinScallop

200024.9028.8826.4819.3540.2942.2540.2538.3740.7242.2541.2538.65
400033.9538.9036.9026.0752.4754.1053.5549.7652.7154.2852.5451.31
600040.8540.5737.4544.5457.9460.7658.2554.8157.7359.2458.5155.43
800042.5145.5142.3439.6761.6363.4961.9359.4662.2264.6062.2659.79
1000044.7250.5844.2939.2965.2768.3966.1161.3164.9167.6065.3361.79
1200049.5348.7548.1151.7468.0471.6467.9564.5367.4969.0468.9864.45
1400048.3750.3550.5944.1670.5072.8670.7467.8970.4772.6470.3568.41
1600054.1253.4050.4858.4973.6176.1774.2370.4472.7275.1673.4569.54
1800052.9559.2955.0844.4875.3878.8474.9972.3174.3276.5175.4071.07
2000057.3358.0454.5459.4277.2080.8777.0573.6976.8278.9178.0473.51
2200058.0457.2159.2457.6779.5482.8079.7076.1378.4880.7279.7075.02
2400055.4462.4459.1444.7581.1885.6180.8077.1480.3782.2882.0376.80
2600057.8561.9862.5249.0583.0286.3182.7080.0683.3485.4684.4880.07
2800060.4262.8162.5655.8985.0188.5285.2581.2584.6987.3286.0480.70
3000059.7467.9161.4749.8585.6089.4485.7881.5886.5489.3387.9282.39
3200062.3265.7363.5257.7284.9088.0685.5681.0687.6590.1588.5584.25
3400063.6068.7265.4456.6484.9688.2885.7980.8287.9290.9689.5583.26
3600064.7072.7665.5555.8085.1388.1185.4281.8787.3489.3688.0884.57
3800063.7369.7165.5055.9885.6589.4085.3382.2187.9089.6289.2884.79
4000071.7770.2667.5277.5485.0288.4685.8380.7687.6989.8888.1785.04
4200067.6271.4669.7861.6385.1289.4784.8980.9987.5890.2787.9984.48
4400069.2571.8266.4569.4784.7987.9184.4781.9987.5790.5688.0884.08
4600066.4271.2671.0256.9985.0489.0184.4581.6687.1589.4888.6483.33
4800066.1970.4666.9661.1585.0588.3785.1081.6687.2189.2788.4283.95
5000070.3368.8670.3571.7884.5988.7384.7880.2787.4290.6987.6883.91