Skip to main content

Table 1 The performance of two models are compared using the metrics, MS COCO's Average Precision (AP) and PASCAL's APx0 (IoU over a threshold of 0.x), and AP (APS, APM, APL) for different object sizes (Small,, Medium, Large)

From: A deep learning method for counting white blood cells in bone marrow images

Models

AP

AP50

AP75

APS

APM

APL

Improved model

0.744

0.853

0.863

0.988

0.831

0.755

Faster RCNN

0.715

0.844

0.838

0.988

0.830

0.725

FPN

0.678

0.800

0.795

0.950

0.826

0.689