Closed rouge012 closed 1 year ago
YuHengsss
commented
1 year ago hello，for AP50 calculation, you could obtain the result following the process of "Reproduce our results on VID" in the readme file.
For time measurement, please refer to https://github.com/YuHengsss/YOLOV/blob/47237acf12ddd6a6173a079b789a0cb38ffb0da7/tools/yolov_demo_online.py#L141 It's worth to mention that this time varies from different CPU because of the post-processing procedure and "TopK" sorting.
Hikaylee
commented
1 year ago hello，for AP50 calculation, you could obtain the result following the process of "Reproduce our results on VID" in the readme file. For time measurement, please refer to
It's worth to mention that this time varies from different CPU because of the post-processing procedure and "TopK" sorting.
您好，我想请问一下为什么用vid_eval和REPPM两种评估的精度差别这么大呢？
YuHengsss
commented
1 year ago hello，for AP50 calculation, you could obtain the result following the process of "Reproduce our results on VID" in the readme file. For time measurement, please refer to https://github.com/YuHengsss/YOLOV/blob/47237acf12ddd6a6173a079b789a0cb38ffb0da7/tools/yolov_demo_online.py#L141
It's worth to mention that this time varies from different CPU because of the post-processing procedure and "TopK" sorting.
您好，我想请问一下为什么用vid_eval和REPPM两种评估的精度差别这么大呢？
应该是 VID AP@50 与 COCO API 不太一样造成的，在我们的测试中，会差大约0.5～0.7个点
Hi! which file did you calculate the AP and Times? I want to reproduce them and calculate other metrics.
Thanks a lot!