How to increase sample size for precision-recall calculation

[holtvogt]

During my evaluation, I came across the predefined sample size several times. I would like to modify it for a more accurate precision-recall graph. How is this possible? Increasing it by hand and providing an additional value into the get_thresholds() method changes the output size, but also leads to a false calculation of AP. I'm referring to

mmdetection3d/mmdet3d/core/evaluation/kitti_utils/eval.py

Line 483 in 962fc83

N_SAMPLE_PTS = 41

Edit: It seems that the get_mAP methods are falsely implemented as they are specifically hardcoded to work with predefined sample size of 41, and thus, aren't adjusting to changing sample sizes:

def get_mAP11(prec):
    sums = 0
    # Magic number 4
    for i in range(0, prec.shape[-1], 4):
        sums = sums + prec[..., i]
    return sums / 11 * 100

def get_mAP40(prec):
    sums = 0
    # Magic number '1' as step size is defaulted to 1
    for i in range(0, prec.shape[-1]):
        sums = sums + prec[..., i]
    return sums / 40 * 100