张量板图召回 [英] Tensorboard graph recall

问题描述

我正在训练一个物体检测器，并运行了评估工作.我在张量板中看到了某些图表.如图所示，tensorflowboard 中的 DetectionBoxes_Recall/AR@10 vs AR@100 vs AR@100(medium) 是什么.DetectionBoxes_Precision/mAP、mAP(大)、mAP(中)、mAP(小)、mAP(0.50IOU)和mAP(0.75IOU)之间有什么区别?请帮助我对此很陌生，谢谢.

I am training an object detector and I ran the evaluation job. I see certain graphs in the tensorboard. What is DetectionBoxes_Recall/AR@10 vs AR@100 vs AR@100(medium) in the tensorflowboard as shown. And what is the difference between DetectionBoxes_Precision/mAP, mAP(large), mAP(medium), mAP(small), mAP(0.50IOU) and mAP(0.75IOU)? Please help I am very new to this thank you.

推荐答案

'DetectionBoxes_Precision/mAP':平均精度在 IOU 阈值范围从 0.5 到 0.95，增量为 0.05 的范围内平均.

'DetectionBoxes_Precision/mAP': mean average precision over classes averaged over IOU thresholds ranging from .5 to .95 with .05 increments.

'DetectionBoxes_Precision/mAP@.50IOU':50% IOU 时的平均精度

'DetectionBoxes_Precision/mAP@.50IOU': mean average precision at 50% IOU

'DetectionBoxes_Precision/mAP@.75IOU':75% IOU 时的平均精度

'DetectionBoxes_Precision/mAP@.75IOU': mean average precision at 75% IOU

'DetectionBoxes_Precision/mAP (small)':小物体(面积 <32^2 像素)的平均精度.

'DetectionBoxes_Precision/mAP (small)': mean average precision for small objects (area < 32^2 pixels).

'DetectionBoxes_Precision/mAP (medium)':中等大小物体的平均精度(32^2 像素 <面积 <96^2 像素).

'DetectionBoxes_Precision/mAP (medium)': mean average precision for medium sized objects (32^2 pixels < area < 96^2 pixels).

'DetectionBoxes_Precision/mAP (large)':大型物体的平均精度(96^2 像素 <面积 <10000^2 像素).

'DetectionBoxes_Precision/mAP (large)': mean average precision for large objects (96^2 pixels < area < 10000^2 pixels).

'DetectionBoxes_Recall/AR@1':1 次检测的平均召回率.

'DetectionBoxes_Recall/AR@1': average recall with 1 detection.

'DetectionBoxes_Recall/AR@10':10 次检测的平均召回率.

'DetectionBoxes_Recall/AR@10': average recall with 10 detections.

'DetectionBoxes_Recall/AR@100':100 次检测的平均召回率.

'DetectionBoxes_Recall/AR@100': average recall with 100 detections.

'DetectionBoxes_Recall/AR@100(小)':100个小物体的平均召回率.

'DetectionBoxes_Recall/AR@100 (small)': average recall for small objects with 100.

'DetectionBoxes_Recall/AR@100 (medium)':中等对象的平均召回率为 100.

'DetectionBoxes_Recall/AR@100 (medium)': average recall for medium objects with 100.

'DetectionBoxes_Recall/AR@100 (large)':100 次检测的大对象的平均召回率.

'DetectionBoxes_Recall/AR@100 (large)': average recall for large objects with 100 detections.

这篇关于张量板图召回的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！