Evaluation on Crowdpose test/val set

[RedOrient]

您好，我在做遮挡评估的时候遇到了一些问题

前提：我的模型加入了3dpw-train set训练而成

我评估了我训练的模型结果如下：

DONE (t=0.22s).
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.267
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.548
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.231
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.480
 Average Recall     (AR) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.781
 Average Recall     (AR) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.500
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=  easy | maxDets= 20 ] = -1.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets= 20 ] = 0.075
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=  hard | maxDets= 20 ] = -1.000

同时评估了ROMP_HRNet32_V1.pkl，结果如下：

DONE (t=0.18s).
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.002
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.004
 Average Recall     (AR) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.020
 Average Recall     (AR) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.001
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=  easy | maxDets= 20 ] = -1.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets= 20 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=  hard | maxDets= 20 ] = -1.000

这个结果和论文中的结果一点也不一样

请问这个结果还需要再进行一步计算才能得到论文中的结果吗？

[Arthur151]

并不需要进一步的计算。 表中数字是Average Precision (AP) @[ IoU=0.50 | area= all | maxDets= 20 ] 这项，可能训练要稍微进行的久一点，checkpoint的选择也会有一点影响。 评估ROMP_HRNet32_V1.pkl，明显有bug。