SwinTransformer / Swin-Transformer-Object-Detection

This is an official implementation for "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows" on Object Detection and Instance Segmentation.
https://arxiv.org/abs/2103.14030
Apache License 2.0
1.81k stars 381 forks source link

Train common backbone with multiple detection heads #194

Open jinga-lala opened 2 years ago

jinga-lala commented 2 years ago

Hi, I want to train COCO and VOC together using the same backbone. How can I ensure that the concat dataset has separate obj detection heads catering to the different number of classes in the two datasets, 80 and 20 respectively. Is there any better workaround? Any help will be appreciated!

Thanks