Xinlong Wang
Xinlong Wang
@LiangYong1216 Can you post your related code? I can help check it.
@Jackyinuo Use this online server: https://competitions.codalab.org/competitions/20796
@shakey-cuimiao Is there enough memory?
@jinfagang 1. Maybe you can try a smaller mask_thr. 2. Did you train the model on your own data? For COCO trained model, we didn't find such results. Maybe you...
@twangnh It's weird. Do you figure it out?
@pvthuy What's your loss functions? I suggest using Focal Loss as mask prediction loss. And did you mean the class scores are low or mask prediction after sigmoid?
@pvthuy If you didn't change the loss part of that config file, then you were using dice loss for mask prediction. However, for your case, Focal Loss might be better....
@Wei-Mao Sorry we currently don't have time to organize the panoptic segmentation part. We may publish those code and models if we are free after the CVPR ddl. But it...
@JarintotionDin The SOLO implementation in mmdet v2 is at [here](https://github.com/WXinlong/mmdetection/tree/SOLO).
Did you train from scratch? If you can provide the config file, I am happy to help check it.