peraktong / Calculate-mAP-for-object-detection

Jason's toy model for calculating mAP

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Calculate mAP for object detection

mAP(mean average precision) is an important parameter to tell whether an object detection model is good. Here I represent a python code to calculate mAP with given boxes edge from your prediction and data. The prediction should be in shape=(-1,5) [probability, xmin, ymin, xmax, ymax]and data in shape=(-1,4) [xmin, ymin, xmax, ymax]. If there is any bug inside, please contact me. Thanks! I didn't include probability(That a box is a real box) into account for now. Coming soon. Remember the data I include in our toy model is not accurate since I use a pre-trained yolov2 model which has different classes from the validation set. This is just a formula to calculate mAP.

About

Jason's toy model for calculating mAP

License:MIT License


Languages

Language:Python 100.0%