基于 Deeplabv3 多尺度训练、多类别分割：路面水坑分割【数据集+代码+训练结果】

[train hyper-parameters: Namespace(base_size=300, batch_size=8, crop_size=224, epochs=5, lr=0.01, lrf=0.001)] [epoch: 1] global correct: 0.8881 precision: ['0.9110', '0.6954', '0.0000'] recall: ['0.9617', '0.4831', '0.0000'] IoU: ['0.8791', '0.3987', '0.0000'] mean IoU: 0.4259 [epoch: 2] global correct: 0.8875 precision: ['0.9495', '0.6120', '0.0000'] recall: ['0.9159', '0.7324', '0.0000'] IoU: ['0.8733', '0.5002', '0.0000'] mean IoU: 0.4578 [epoch: 3] global correct: 0.9108 precision: ['0.9309', '0.7656', '0.0000'] recall: ['0.9664', '0.6055', '0.0000'] IoU: ['0.9017', '0.5108', '0.0000'] mean IoU: 0.4708 [epoch: 4] global correct: 0.9151 precision: ['0.9468', '0.7323', '0.0000'] recall: ['0.9532', '0.7061', '0.0000'] IoU: ['0.9048', '0.5612', '0.0000'] mean IoU: 0.4887 [epoch: 5] global correct: 0.9171 precision: ['0.9495', '0.7345', '0.0000'] recall: ['0.9527', '0.7221', '0.0000'] IoU: ['0.9068', '0.5726', '0.0000'] mean IoU: 0.4932