ZhangHaipeng Deep Learning, Software Engineer

2024-08-16

accuracy, recall, precision, confusion matrix F1-score, mAP

在谈到分类评价问题是, 应首先要了解下几个基本概念,基本上所有的评价指标都是围绕着这几个 基本概念定义的:

  • TP, True Positive 真阳性: 实际为正,预测为正
  • FP, False Positive 假阳性: 实际为负,预测为正
  • FN, False Negative 假阴性: 实际为正,预测为负
  • TN,True Negative 真阴性: 实际为负,预测为负

P = TP + FN N = TN + FP 正常情况: Accuracy = (TP + TF) / (P + N)


Comments

Content