用PaddleDetection训练烟雾火灾数据集，采用默认参数，25个epoch，损失20多，精度10%不到，哪里出了问题？

Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.025
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.097
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.007
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.011
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.031
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.030
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.035
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.121
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.169
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.109
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.128
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.178
[03/11 14:28:25] ppdet.engine INFO: Total sample number: 333, averge FPS: 28.076371977477603
[03/11 14:28:25] ppdet.engine INFO: Best test bbox ap is 0.025.
[03/11 14:28:26] ppdet.utils.checkpoint INFO: Save checkpoint: output/yolov3_mobilenet_v1_270e_coco
[03/11 14:28:26] ppdet.engine INFO: Epoch: [25] [ 0/375] learning_rate: 0.001000 loss_xy: 7.372954 loss_wh: 2.801722 loss_obj: 16.380930 loss_cls: 0.681809 loss: 27.227947 eta: 2:20:34 batch_cost: 0.8777 data_cost: 0.7679 ips: 9.1142 images/s
[03/11 14:31:40] ppdet.engine INFO: Epoch: [25] [200/375] learning_rate: 0.001000 loss_xy: 7.599291 loss_wh: 2.757879 loss_obj: 17.024263 loss_cls: 0.617985 loss: 28.201618 eta: 2:17:47 batch_cost: 0.9647 data_cost: 0.8496 ips: 8.2927 images/s
[03/11 14:34:30] ppdet.engine INFO: Epoch: [26] [ 0/375] learning_rate: 0.001000 loss_xy: 7.299512 loss_wh: 2.797149 loss_obj: 16.613195 loss_cls: 0.686537 loss: 27.622408 eta: 2:15:20 batch_cost: 0.9765 data_cost: 0.8653 ips: 8.1927 images/s
[03/11 14:37:45] ppdet.engine INFO: Epoch: [26] [200/375] learning_rate: 0.001000 loss_xy: 7.076797 loss_wh: 2.615398 loss_obj: 15.749668 loss_cls: 0.650539 loss: 26.118111 eta: 2:12:31 batch_cost: 0.9719 data_cost: 0.8524 ips: 8.2311 images/s
[03/11 14:40:32] ppdet.engine INFO: Epoch: [27] [ 0/375] learning_rate: 0.001000 loss_xy: 7.079313 loss_wh: 2.631054 loss_obj: 16.041780 loss_cls: 0.678031 loss: 27.228664 eta: 2:10:00 batch_cost: 0.9453 data_cost: 0.8287 ips: 8.4626 images/s
[03/11 14:43:33] ppdet.engine INFO: Epoch: [27] [200/375] learning_rate: 0.001000 loss_xy: 7.418595 loss_wh: 2.738465 loss_obj: 16.423649 loss_cls: 0.763486 loss: 27.510603 eta: 2:06:59 batch_cost: 0.9013 data_cost: 0.7872 ips: 8.8760 images/s