Feng Wang
Evaluator supports logging recall value now, you could use Tensorboard to log it.
@improgress You could get AP/AR like code [here](https://github.com/Megvii-BaseDetection/YOLOX/blob/main/yolox/evaluators/coco_evaluator.py#L285-L290)
> My dataset format is Voc ,and I don’t see some information about AR in voc_evaluator.py. @improgress AR of VOC is not supported now.
It depends on the max annoation number of a single image in your dataset. If it's greater than config in YOLOX, plz set the value.
Thanks @glenn-jocher for inviting us to attend the conference. It sounds interesting and amazing. Unfortunately, we are all busy these days and couldn't attend the conference. BTW, If the community...
It' s simple if you stop your process by using `ctrl + c`. Just kill the left process with `kill` command.
No, I mean if there exists something like GPU memory not released totally after using `ctrl + c` to stop a process, just using `kill` command to kill the process....
Sorry, I could not understand the exact problem you are facing, could you explain it in detail? Any exception backtrack of ` process 1 terminated with signal SIGABRT` ?
@nahidalam For a long time training, RGB and BGR doesn't matter too much. I think fix bug in `augment_hsv` could help, but fixing this might not make the result better.
Could you post your error log with more information?