Swin-Transformer
Swin-Transformer copied to clipboard
microsoft
This is an official implementation for "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows".
I think your Swin-MLP code is different from Swin-Mixer in your paper, is it right? And if you can, can you explain how did you apply it to MLP Mixer...
In `/Swin-Transformer-Object-Detection/tools/test.py`,there is `evaluation metrics, which depends on the dataset, e.g., "bbox",' ' "segm", "proposal" for COCO, and "mAP", "recall" for PASCAL VOC'`. And I know that bbox is for...
I train the model with img_size = 112 ``` SwinTransformer(img_size=112, patch_size=4, num_classes=512, embed_dim=96, depths=[2, 2, 6, 2], num_heads=[3, 6, 12, 24], window_size=7, drop_path_rate=0.2, use_checkpoint=False) ``` and error occurs ``` "swin_transformer.py",...
I wonder whether Swin Transformer can directly take images with different size without any resizing preprocessing?
How to properly upscale SWIN network to 1 mega pixel? There are two configurations in my mind: _Configuration1:_ DATA: IMG_SIZE: 1024 MODEL: TYPE: swin NAME: swin_small_patch4_window7_1024 SWIN: EMBED_DIM: 96 DEPTHS:...
我尝试着自己手搭Swin做语义分割。首先我把Swin的每一步输出做view()操作，然后再通过一个上采样部分，用的转置卷积上采样。但是模型训练结果很差劲，同一张图片只有一个颜色。我不知道自己问题出在哪里了，请问对Swin的输出直接view()然后去上采样合适嘛？
Thanks for your great work! We applied Swin Transformer to video frame interpolation. With good efficiency, the model achieves the state-of-the-art performance on video interpolation on several benchmark datasets, including...
If I pre-train Swin-T for 224 input image size. How can I finetune it and get Swin-T for 320 input image size? In your paper, you claimed 384^2 input models...
I'm very interested in training giant model. Especially, trying to reproduce Swin-V2-G. Does MS-Asia will have plan to release Swin-V2 codes? Or could you guide me to how to impl...
