什么是 mAP 指标，它是如何计算的? [英] What is the mAP metric and how is it calculated?

问题描述

在计算机视觉和物体检测中，常用的评估方法是mAP.它是什么，如何计算?

In computer vision and object detection, the common evaluation method is mAP. What is it and how is it calculated?

推荐答案

引用来自上述Zisserman 论文 - 4.2 结果评估(第 11 页):

首先，重叠标准"定义为大于 0.5 的交集.(例如，如果一个预测框满足这个关于真实框的标准，它被认为是一个检测).然后使用这种贪婪"方法在 GT 框和预测框之间进行匹配:

First an "overlap criterion" is defined as an intersection-over-union greater than 0.5. (e.g. if a predicted box satisfies this criterion with respect to a ground-truth box, it is considered a detection). Then a matching is made between the GT boxes and the predicted boxes using this "greedy" approach:

一种方法输出的检测被分配给ground truth对象按(递减)排序的顺序满足重叠标准信心输出.图像中同一对象的多次检测被认为是错误检测，例如5 次检测单个物体计为 1 次正确检测和 4 次错误检测

Detections output by a method were assigned to ground truth objects satisfying the overlap criterion in order ranked by the (decreasing) confidence output. Multiple detections of the same object in an image were considered false detections e.g. 5 detections of a single object counted as 1 correct detection and 4 false detections

因此每个预测框要么是真阳性，要么是假阳性.每个ground-truth box都是True-Positive.没有真正的否定.

Hence each predicted box is either True-Positive or False-Positive. Each ground-truth box is True-Positive. There are no True-Negatives.

然后通过对精确召回曲线上的精确度值进行平均来计算平均精确度，其中召回率在 [0, 0.1, ..., 1] 范围内(例如，11 个精确度值的平均值).更准确地说，我们考虑一个稍微修正的 PR 曲线，其中对于每个曲线点 (p, r)，如果存在不同的曲线点 (p', r') 使得 p' > p 并且 r' >= r，我们将 p 替换为这些点中的最大 p'.

Then the average precision is computed by averaging the precision values on the precision-recall curve where the recall is in the range [0, 0.1, ..., 1] (e.g. average of 11 precision values). To be more precise, we consider a slightly corrected PR curve, where for each curve point (p, r), if there is a different curve point (p', r') such that p' > p and r' >= r, we replace p with maximum p' of those points.

我仍然不清楚的是，对于那些从未检测到的 GT 框做了什么(即使置信度为 0).这意味着精度-召回曲线永远无法达到某些召回值，这使得上述平均精度计算未定义.

What is still unclear to me is what is done with those GT boxes that are never detected (even if the confidence is 0). This means that there are certain recall values that the precision-recall curve will never reach, and this makes the average precision computation above undefined.

简答:在recall不可达的区域，精度下降到0.

Short answer: in the region where the recall is unreachable, the precision drops to 0.

解释这一点的一种方法是假设当置信度的阈值接近 0 时，无限数量的预测边界框会在整个图像中亮起.然后精度立即变为 0(因为只有有限数量的 GT 框)，并且召回率在这条平坦曲线上不断增长，直到达到 100%.

One way to explain this is to assume that when the threshold for the confidence approaches 0, an infinite number of predicted bounding boxes light up all over the image. The precision then immediately goes to 0 (since there is only a finite number of GT boxes) and the recall keeps growing on this flat curve until we reach 100%.

这篇关于什么是 mAP 指标，它是如何计算的?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！