在训练detectron2的maskrcnn模型的验证过程中出现WARNING [04/03 13:19:54 d2.evaluation.coco_evaluation]: No predictions from the model!， 大意是模型没有输出，自己尝试用模型测试图片，发现随着epoch增加，锚框的数量越来越少，最初是有一些，但是精度很差，往后就根本没有了，全是background，即使对锚框置信度要求低也没有。
训练过程中出现的warning：
[04/03 13:01:50 d2go.data.build]: Using dataset mapper:
D2GoDatasetMapper:
is_train: True
image_loader: None
tfm_gens:
- ResizeShortestEdge(short_edge_length=(224,), max_size=320, sample_style='choice')
- RandomFlip()
WARNING [04/03 13:01:50 d2.data.datasets.coco]:
Category ids in annotations are not in [1, #categories]! We'll apply a mapping for you.
[04/03 13:19:12 d2go.runner.default_runner]: Using dataset mapper:
D2GoDatasetMapper:
is_train: False
image_loader: None
tfm_gens:
- ResizeShortestEdge(short_edge_length=(224, 224), max_size=320, sample_style='choice')
WARNING [04/03 13:19:13 d2.data.datasets.coco]:
Category ids in annotations are not in [1, #categories]! We'll apply a mapping for you.
[04/03 13:19:13 d2.data.datasets.coco]: Loaded 1311 images in COCO format from /home/mycf/桌面/Objs_Table/annotations.json
[04/03 13:19:13 d2.data.common]: Serializing 1311 elements to byte tensors and concatenating them all ...
[04/03 13:19:13 d2.data.common]: Serialized dataset takes 6.50 MiB
[04/03 13:19:13 d2go.runner.default_runner]: Adding visualization evaluator ...
WARNING [04/03 13:19:13 d2.data.datasets.coco]:
Category ids in annotations are not in [1, #categories]! We'll apply a mapping for you.