how the paper calculates the AP and mAP?

#95
by chenxiangyi10 - opened

The paper mentions that the metric of standard average precision is used. However, how to get the value of this metric is unclear. Getting the precision-recall curve for the models with confidence scores of bounding boxes is pretty straightforward. Do you use the confidence score generated by Florence 2 to obtain the AP and mAP?

Sign up or log in to comment