Closed pingjun18-li closed 3 years ago
We use the ImageNet "mean" and "std".
I plan to release a new codebase that includes models from 5 semantic segmentation datasets (including Mapillary Vistas) and 2 panoptic segmentation datasets. I will let you know when it is online. Stay tuned!
ok, thank you so much for your great jobs! And looking forward to your new achievements!
@pingjun18-li
Please check my new codebase for semantic and panoptic segmentation: https://github.com/facebookresearch/MaskFormer
It supports (and have pre-trained weights) 5 major semantic segmentation datasets and 2 panoptic segmentation datsaets.
Wow, nice, I'll take it seriously and make full use of it!
Sorry to bother you again. I want to do an experiment on MV(Mapillary Vistas) dataset, but I have found many projects without its description and haven't shared the codes about training on MV. Would you like to share the value of "mean" and "std" about MV? Thank you so much!