X-Decoder
X-Decoder copied to clipboard
microsoft
Official Implementation of X-Decoder for generalized decoding for pixel, image and language
In your paper you report the mIoU of the X-Decoder (T) model to be 96.2. I tried to reproduce these results. I did not find the appropriate evaluation script so...
In this line (below), it seems that the code uses `masks` to predict boxes, https://github.com/microsoft/X-Decoder/blob/165f8a6314ac84f5c36aaab7216f90dd97e38a43/modeling/architectures/xdecoder_model.py#L922 but in [line 913](https://github.com/microsoft/X-Decoder/blob/165f8a6314ac84f5c36aaab7216f90dd97e38a43/modeling/architectures/xdecoder_model.py#L913), the predicted boxes are already obtained. https://github.com/microsoft/X-Decoder/blob/165f8a6314ac84f5c36aaab7216f90dd97e38a43/modeling/architectures/xdecoder_model.py#L913 Why do not use...
I tried both the 2.0 and main branches , and produced the same error. Part of the error code: ` class JointLoader(torchdata.IterableDataset): def __init__(self, loaders, key_dataset): dataset_names = [] for...
why are "vlp_train" and "vlp_val" same? https://github.com/microsoft/X-Decoder/blob/165f8a6314ac84f5c36aaab7216f90dd97e38a43/datasets/registration/register_vlp_datasets.py#L27 https://github.com/microsoft/X-Decoder/blob/165f8a6314ac84f5c36aaab7216f90dd97e38a43/datasets/registration/register_vlp_datasets.py#L22
When I use the demo demo on huggingface for image segmentation, I get great results：  But when I use the published pre-trained model for segmentation, I get poor results：...
Hello, I'd like to evalute bdd datasets, but there are some errors in the code. I wonder where to get the below files? "bdd100k/labels/pan_seg/coco_pano/val", "bdd100k/labels/pan_seg/meta/coco_val.json"
I did not find where the evaluation code exist for VQA and Interactive in X-decoder
Hi! Thanks for this amazing work! I'm having a hard time evaluating on VOC2012 for semantic segmentation. Could you provide clear instructions how to reproduce the numbers reported in your...
Hi, thanks for releasing the code. Why do you choose `MPI adaptor` instead of using default `torch.distributed` (just like in [Mask2Former](https://github.com/facebookresearch/Mask2Former/blob/9b0651c6c1d5b3af2e6da0589b719c514ec0d69a/train_net.py#L321C8-L321C8)) ? ```[tasklist] ### Tasks ```
Thank you for your great work. I am trying to run the X-Decoder demo in a local environment. I want to run run_demo.sh but it doesn't work because torch and...
