NVlabs/SegFormer issues and pull requests

#158 - How to get the pretrained weight!

Issue - State: open - Opened by ywxhjn about 1 month ago

#151 - the pre-trained weight of Mit b0 - b5

Issue - State: open - Opened by xiuguangLi 4 months ago - 2 comments

#100 - How the final output is converted to full resolution mask?

Issue - State: open - Opened by working12 about 2 years ago

#100 - How the final output is converted to full resolution mask?

Issue - State: open - Opened by working12 about 2 years ago

#99 - xxx is not a checkpoint file

Issue - State: open - Opened by toohappygirl about 2 years ago

#99 - xxx is not a checkpoint file

Issue - State: open - Opened by toohappygirl about 2 years ago

#98 - About the PatchEmbedding

Issue - State: open - Opened by SimonServant about 2 years ago - 2 comments

#97 - imagenet train code

Issue - State: closed - Opened by endaoguansanlu over 2 years ago - 1 comment

#96 - torch onnx export

Issue - State: open - Opened by germanjke over 2 years ago - 1 comment

#95 - Missing dependencies

Issue - State: open - Opened by Tanmay-Kulkarni101 over 2 years ago

#94 - KeyError: "EncoderDecoder: 'mit_b1 is not in the models registry'"

Issue - State: open - Opened by endaoguansanlu over 2 years ago - 3 comments

#93 - assert palette.shape[0] == len(self.CLASSES)

Issue - State: open - Opened by AhmadZobairSurosh over 2 years ago - 5 comments

#92 - Training On less than 150 num_classes, triggers Cuda error

Issue - State: closed - Opened by AhmadZobairSurosh over 2 years ago - 1 comment

#92 - Training On less than 150 num_classes, triggers Cuda error

Issue - State: closed - Opened by AhmadZobairSurosh over 2 years ago - 1 comment

#91 - Result per class returns nan

Issue - State: open - Opened by AhmadZobairSurosh over 2 years ago - 1 comment

#91 - Result per class returns nan

Issue - State: open - Opened by AhmadZobairSurosh over 2 years ago - 1 comment

#90 - Bad github link for mmsegmentation in Dockerfile

Issue - State: open - Opened by pavllan90 over 2 years ago

#89 - About the result of MLP decoder

Issue - State: open - Opened by ShuGuoJ over 2 years ago

#89 - About the result of MLP decoder

Issue - State: open - Opened by ShuGuoJ over 2 years ago

#88 - 论文中的figure 2

Issue - State: open - Opened by tricksterANDthug over 2 years ago

#88 - 论文中的figure 2

Issue - State: open - Opened by tricksterANDthug over 2 years ago

#87 - How can i convert the weights to tensorrt engine file?

Issue - State: open - Opened by JiangWeiHn over 2 years ago - 1 comment

#87 - How can i convert the weights to tensorrt engine file?

Issue - State: open - Opened by JiangWeiHn over 2 years ago - 1 comment

#86 - Add in timm ?

Issue - State: open - Opened by jiarenyf over 2 years ago

#86 - Add in timm ?

Issue - State: open - Opened by jiarenyf over 2 years ago

#85 - What should be the next step to compute point-to-point predictions after getting the (H/4, W/4) mask? Should it simply apply an interpolation on the (H/4, W/4) mask to obtain a prediction map of original size of (H, W)?

Issue - State: open - Opened by SudongCAI over 2 years ago - 2 comments

#85 - What should be the next step to compute point-to-point predictions after getting the (H/4, W/4) mask? Should it simply apply an interpolation on the (H/4, W/4) mask to obtain a prediction map of original size of (H, W)?

Issue - State: open - Opened by SudongCAI over 2 years ago - 2 comments

#84 - What should be the next step to compute point-to-point predictions after getting the (H/4, W/4) mask? Should it simply apply an interpolation on the (H/4, W/4) mask to obtain a prediction map of original size of (H, W)?

Issue - State: closed - Opened by SudongCAI over 2 years ago

#84 - What should be the next step to compute point-to-point predictions after getting the (H/4, W/4) mask? Should it simply apply an interpolation on the (H/4, W/4) mask to obtain a prediction map of original size of (H, W)?

Issue - State: closed - Opened by SudongCAI over 2 years ago

#83 - MMSegmentation Tutorial.ipynb fails when running Google Colab

Issue - State: open - Opened by chris-kehl over 2 years ago - 1 comment

#82 - A question about train on single GPU

Issue - State: closed - Opened by xibi777 over 2 years ago - 1 comment

#82 - A question about train on single GPU

Issue - State: closed - Opened by xibi777 over 2 years ago - 1 comment

#81 - Sorry, None problem.

Issue - State: closed - Opened by winycg over 2 years ago

#81 - Sorry, None problem.

Issue - State: closed - Opened by winycg over 2 years ago

#80 - A question about the parameters and size

Issue - State: open - Opened by Leon-Lee-WC over 2 years ago

#80 - A question about the parameters and size

Issue - State: open - Opened by Leon-Lee-WC over 2 years ago

#79 - what configuration are you using in win10

Issue - State: closed - Opened by stu-yzZ over 2 years ago

#78 - custom dataset training for segmentation

Issue - State: open - Opened by soheilmajidi over 2 years ago - 3 comments

#78 - custom dataset training for segmentation

Issue - State: open - Opened by soheilmajidi over 2 years ago - 3 comments

#77 - HOW to reproduce Fig. 3?

Issue - State: open - Opened by fingertap over 2 years ago

#77 - HOW to reproduce Fig. 3?

Issue - State: open - Opened by fingertap over 2 years ago

#76 - the segformer not converge on my custom dataset

Issue - State: open - Opened by AncientRemember over 2 years ago - 2 comments

#76 - the segformer not converge on my custom dataset

Issue - State: open - Opened by AncientRemember over 2 years ago - 2 comments

#75 - SegFormer for binary semantic segmentation problem

Issue - State: open - Opened by loucif01 over 2 years ago - 5 comments

#75 - SegFormer for binary semantic segmentation problem

Issue - State: open - Opened by loucif01 over 2 years ago - 5 comments

#74 - SegFormer 3D

Issue - State: open - Opened by sulaimanvesal over 2 years ago - 1 comment

#74 - SegFormer 3D

Issue - State: open - Opened by sulaimanvesal over 2 years ago - 1 comment

#73 - Bucket size warning

Issue - State: open - Opened by rahulagrawal048 over 2 years ago - 5 comments

#72 - how to train and test SegFormer on my own dateset..

Issue - State: open - Opened by ZhuangChen25674 over 2 years ago - 4 comments

#72 - how to train and test SegFormer on my own dateset..

Issue - State: open - Opened by ZhuangChen25674 over 2 years ago - 4 comments

#71 - robustness test

Issue - State: open - Opened by wwjwy over 2 years ago

#70 - Config and Dataloader files for ImageNet pre-training

Issue - State: open - Opened by Midhun07 over 2 years ago

#70 - Config and Dataloader files for ImageNet pre-training

Issue - State: open - Opened by Midhun07 over 2 years ago

#69 - Problem with loading models

Issue - State: open - Opened by Ngheissari almost 3 years ago - 1 comment

#68 - No module named 'mmseg'

Issue - State: closed - Opened by ooodragon94 almost 3 years ago - 5 comments

#68 - No module named 'mmseg'

Issue - State: closed - Opened by ooodragon94 almost 3 years ago - 5 comments

#67 - Pre-train Code or implementation details

Issue - State: closed - Opened by yi94code almost 3 years ago - 1 comment

#67 - Pre-train Code or implementation details

Issue - State: closed - Opened by yi94code almost 3 years ago - 1 comment

#66 - fixed a misspelling in Dockerfile

Pull Request - State: closed - Opened by yamato-roadroller almost 3 years ago

#66 - fixed a misspelling in Dockerfile

Pull Request - State: closed - Opened by yamato-roadroller almost 3 years ago

#65 - Add convolution block before the entire model

Issue - State: open - Opened by Shravan-3501 almost 3 years ago - 1 comment

#65 - Add convolution block before the entire model

Issue - State: open - Opened by Shravan-3501 almost 3 years ago - 1 comment

#64 - Computing the effective receptive fields (figure 6 of the paper)

Issue - State: closed - Opened by ArthurZucker almost 3 years ago - 3 comments

#64 - Computing the effective receptive fields (figure 6 of the paper)

Issue - State: closed - Opened by ArthurZucker almost 3 years ago - 3 comments

#63 - implementation details regarding Ei : the expansion ratio

Issue - State: open - Opened by DarrenIm almost 3 years ago - 2 comments

#62 - 关于预训练

Issue - State: open - Opened by xiaoachen98 almost 3 years ago

#62 - 关于预训练

Issue - State: open - Opened by xiaoachen98 almost 3 years ago

#61 - gpu

Issue - State: open - Opened by roar-1128 almost 3 years ago

#61 - gpu

Issue - State: open - Opened by roar-1128 almost 3 years ago

#60 - SegFormer for 3D segmentation, and landmark detection

Issue - State: open - Opened by m-ali-awan almost 3 years ago

#59 - Question about the first OverlapPatchEmbed

Issue - State: open - Opened by pieteroyens about 3 years ago - 2 comments

#59 - Question about the first OverlapPatchEmbed

Issue - State: open - Opened by pieteroyens about 3 years ago - 2 comments

#58 - Inference resolution size

Issue - State: open - Opened by ashutosh1807 about 3 years ago - 2 comments

#58 - Inference resolution size

Issue - State: open - Opened by ashutosh1807 about 3 years ago - 2 comments

#57 - The dataset about ADE20K.

Issue - State: open - Opened by Med-Process about 3 years ago - 1 comment

#56 - Cannot use Apex to backward the loss generated by SegFormer

Issue - State: open - Opened by QIU023 about 3 years ago - 1 comment

#55 - 训练SegFormer-B1在cityscapes数据集，出现错误“Defaule process group is not initialized”

Issue - State: open - Opened by znshi about 3 years ago - 8 comments

#55 - 训练SegFormer-B1在cityscapes数据集，出现错误“Defaule process group is not initialized”

Issue - State: open - Opened by znshi about 3 years ago - 8 comments

#54 - Slightly lower performance with ade20k and b0 model

Issue - State: open - Opened by wzpscott about 3 years ago - 3 comments

#53 - The mIoU on the cityscapes dataset is low

Issue - State: open - Opened by fyn111 about 3 years ago - 5 comments

#52 - Question about normalization (Mean/Std) value different from swin pretrained backbone

Issue - State: closed - Opened by MaureenZOU about 3 years ago - 1 comment

#52 - Question about normalization (Mean/Std) value different from swin pretrained backbone

Issue - State: closed - Opened by MaureenZOU about 3 years ago - 1 comment

#51 - The memory usage is high?

Issue - State: open - Opened by feiyuhuahuo about 3 years ago

#51 - The memory usage is high?

Issue - State: open - Opened by feiyuhuahuo about 3 years ago

#50 - Do these functions be used in training?

Issue - State: open - Opened by feiyuhuahuo about 3 years ago - 1 comment

#50 - Do these functions be used in training?

Issue - State: open - Opened by feiyuhuahuo about 3 years ago - 1 comment

#49 - How to reduce the dataset?

Issue - State: open - Opened by peterlv666 about 3 years ago

#48 - How can I fix this error？Thanks!

Issue - State: open - Opened by kekexixili over 3 years ago - 2 comments

#48 - How can I fix this error？Thanks!

Issue - State: open - Opened by kekexixili over 3 years ago - 2 comments

#47 - Can you share the config file for training COCO-stuff

Issue - State: open - Opened by fereenwong over 3 years ago - 5 comments

#46 - Issue in the beginning

Issue - State: open - Opened by bonjour-npy over 3 years ago - 2 comments

#45 - 无法通过预训练模型完成训练

Issue - State: open - Opened by heblade over 3 years ago - 5 comments

#45 - 无法通过预训练模型完成训练

Issue - State: open - Opened by heblade over 3 years ago - 5 comments

#44 - Evaluation报错无法执行成功

Issue - State: open - Opened by heblade over 3 years ago

#44 - Evaluation报错无法执行成功

Issue - State: open - Opened by heblade over 3 years ago

#43 - Can you use some urls to release your pretrain weights?

Issue - State: closed - Opened by clownrat6 over 3 years ago

#43 - Can you use some urls to release your pretrain weights?

Issue - State: closed - Opened by clownrat6 over 3 years ago

#42 - Segformer for medical image segmentation

Issue - State: closed - Opened by qdd1234 over 3 years ago - 1 comment

#41 - How to change checkpoint saving frequency

Issue - State: closed - Opened by gcilli over 3 years ago - 2 comments

#41 - How to change checkpoint saving frequency

Issue - State: closed - Opened by gcilli over 3 years ago - 2 comments

GitHub / NVlabs/SegFormer issues and pull requests