我在训练和评估Open Images数据集时遇到问题,因为没有定义“ oid_challenge_object_detection_metrics ” .

它在文档中定义:https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/evaluation_protocols.md

但是, eval_util.py 中不存在代码,该代码仅定义"coco_detection_metrics"和"coco_mask_metrics"的处理逻辑 .

因此,当我运行object_detection / model_main.py时,它会以错误终止:

File "object_detection/eval_util.py", line 650, in get_eval_metric_ops_for_evaluators
    'Found {} in the evaluation metrics'.format(metric))
ValueError: The only evaluation metrics supported are "coco_detection_metrics" and "coco_mask_metrics". Found oid_challenge_object_detection_metrics in the evaluation metrics

它在我的管道配置中定义:

eval_config: {
  metrics_set: "oid_challenge_object_detection_metrics"
  num_examples: 8000
  max_evals: 10
}

我在 metrics/oid_od_challenge_evaluation.py 下找到了一个文件,但我不确定如何将它集成到eval_util.py中 .

如何正确运行模型培训和评估?谢谢 .