Evaluate errors in mAP for each category

[SCsleepy]

hi，
There are two category in my dataset，Two category will appear simultaneously in each image. When two category are evaluated together, the mAP is 0.609046. When I evaluate category 1 alone, the mAP is 0.429362, and when I evaluate category 2 alone, the mAP is 0.189188. I feel that the result is not quite right. May I ask what the reason is?
I refer to the method of evaluating category separately #2081 .

I hope to receive your clarification.