DiffusionDet
DiffusionDet copied to clipboard
[ICCV2023 Oral] PyTorch implementation of DiffusionDet (https://arxiv.org/abs/2211.09788)
In a common manner,it usually split the dataset to train/val/test three parts. so if i want to use diffusiondet,should i merge the val and test images in one folder?
Great project, thank you to the author for their excellent work. 1.coco `ln -s /path_to_coco_dataset/train2017 datasets/coco/train2017` `ln -s /path_to_coco_dataset/val2017 datasets/coco/val2017` Does train2017 here refer to images or labels? 2. lvis...
Hi, @ShoufaChen thank you for the great work. I am a starter regarding diffusion models, and have a question about scale factor applied in `prepare_diffusion_concat()` and `ddim_sample()` I understood that...
Hi, thanks for your great work. If you freeze the image_encoder (resnet50, swin) during the training stage?
Dear Shoufa, Thank you for the innovative paper on using diffusion to solve the detection problem! I just have one question on the diffusion process proposed in the paper: When...
Thanks for your excellent and inspiring work. I have something confusing. As we all know, after the diffusion process, the common diffusion model like DDPM would use the posterior distribution...
If we add something of our own to the model, which one should we modify if we add loss? Should it be SetCriterionDynamicK or HungarianMatcherDynamicK
assert (boxes1[:, 2:] >= boxes1[:, :2]).all() AssertionError
https://colab.research.google.com/drive/1MIBszjvuBd18ccS3CvczTpXSrTLMwFW6#scrollTo=tudpVdlLnxnw hello, this is my source code. After i register my dataset, the CLI always returns "dataset is not registered". Can you help me please!!