ViTAE-Transformer / ViTDet

Unofficial implementation for [ECCV'22] "Exploring Plain Vision Transformer Backbones for Object Detection"

ViTAE-Transformer/ViTDet Issues

Runtime failure with customize dataset
Updated a year ago1
Adjusting image size
Updated a year ago1
size mismatch and AssertionError
Updated a year ago3
Why do I adjust my img_size to (400,600) and report the following error? QAQ
Updated a year ago1
Welcome update to OpenMMLab 2.0
Updated 2 years ago
when use norm_cfg, there is an error
Updated 2 years ago
How does mmdet training on TPUs ?
Updated 2 years ago
Problems in the use of pre training model
Updated 2 years ago5
[Question] MAE vitdet interval
Updated 2 years ago
ViTAE-B pretrained model does not matched with ViTDet-ViTAE-Base-100e.py?
Updated 2 years ago
Why learning rate of ViTDet is much lower (1e-6) than that in setting (1e-4)?
Closed 2 years ago2
Why loading official MAE pretrained model does not match?
Closed 2 years ago2
"num_classes" error when training customize dataset
Closed 2 years ago2
master_port is already used
Updated 2 years ago1
Why loading pre-trained models does not match？
Updated 2 years ago2
Does the VitDet still adopt FPN rather than the simple pyramid structure in VITDet paper?
Updated 2 years ago2
The model converges slowly during training
Updated 2 years ago
input must be contiguous
Updated 2 years ago2
Input shape error
Closed 2 years ago5
INDEXERROR
Updated 2 years ago3
Have you tried other self-supervised pretrained ViTs like DINO?
Updated 2 years ago1
Getting RuntimeError
Closed 3 years ago12
Environment Setting
Closed 3 years ago2
Details about MAE pre-trained ViTAE
Closed 3 years ago4
Question of PCM
Closed 3 years ago2
How long does it take to train?
Closed 3 years ago2
about lsj_100e config
Closed 3 years ago2