## PASCAL VOC mAP

PASCAL VOC数据集的AP计算经历了两个版本，参考

mAP就是多个AP的平均值

### 版本二：12

$AP = \int _{0}^{1} p_{smooth}(r)dr$

### TP

Multiple detections of the same object in an image are considered false detections e.g. 5 detections of a single object is counted as 1 correct detection and 4 false detections - it is the responsibility of the participant's system to filter multiple detections from its output.

## Precision/Recall

$Precisoin = \frac {TP}{TP+FP} \ Recall = \frac {TP}{TP+FN}$

1. 置信度阈值：判断预测边界框是否为positive
2. IoU：判断预测边界框是否为true/false

• 此时$$M$$个候选边界框中有$$K$$个大于等于置信度阈值的候选边界框
• K个候选边界框中存在$$H$$TP，那么$$FP = K-H$$

$Precision = \frac {H}{K} \ Recall = \frac {H}{N}$

## 实现

PASCAL VOC提供的是Matlab实现，不懂，不过幸好网上有人开源了相应的Python实现