High-Accuracy Real-Time Fish Detection Based on Self-Build Dataset and RIRD-YOLOv3

<table class="table-group" id="tab2"><tr><td><table class="table"><tr><td class="thead-hr" colspan="2"><hr/></td></tr><tr><td class="align_left">Number of epochs</td><td class="align_center">1000, 2000, …, 40000</td></tr><tr><td class="align_left" colspan="2"><hr/></td></tr><tr><td class="align_left">Momentum</td><td class="align_center">0.6, 0.7, 0.8, 0.9</td></tr><tr><td class="align_left" colspan="2"><hr/></td></tr><tr><td class="align_left">Learning rate</td><td class="align_center">0.00001, 0.00002, 0.00003, 0.00004</td></tr><tr><td class="align_left" colspan="2"><hr/></td></tr><tr><td class="align_left">Batch size</td><td class="align_center">8, 16, 32, 64</td></tr><tr class="table-tr"><td colspan="2"><hr class="tbody-hr"/></td></tr></table></td></tr></table>

<div>Tested values in grid search of hyperparameters.</div>

Complexity

tab2

Table 2

Table 2: High-Accuracy Real-Time Fish Detection Based on Self-Build Dataset and RIRD-YOLOv3 