search

WongKinYiu / yolov7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
GNU General Public License v3.0
13.27k stars 4.19k forks source link

Question : What is box mAP in evaluation #1785

Open dinushazoomi opened 1 year ago

dinushazoomi commented 1 year ago

According to the State of the art model evaluation in papers with code, Transformer based object detectors provide better box mAP that yolov7. Can someone pleace give an small explenation on what is this box mAP

alexandrefch commented 1 year ago

Mean Average Precision (mAP) is used to measure the performance of computer vision models. mAP is equal to the average of the Average Precision metric across all classes in a model. You can use mAP to compare both different models on the same task and different versions of the same model. mAP is measured between 0 and 1.

In addition, Average Precision (AP) is compute using the Confusion Matrix.

You can find more information here : https://blog.roboflow.com/mean-average-precision/

dinushazoomi commented 1 year ago

@alexandrefch Thanks for your explanation. Are mAP and box mAP is same or different?

alexandrefch commented 1 year ago

@alexandrefch Thanks for your explanation. Are mAP and box mAP is same or different?

As far as I know, there is no difference, they are the same thing.

jordanesikati commented 1 year ago

Box mAP or mAP(mean average precision) in the context, refers to the bounding box detection metric for the validation phase. It is a way to evaluate and compare object detector models in terms of detection performance. It is to be considered that the same metric can also be applied to other computer vision tasks such as human pose estimation, and instance segmentation.

free-soellingeraj commented 4 months ago

Yes, but at what IOU thresholds are AP calculated?