SwinTransformer/Swin-Transformer-Object-Detection issues and pull requests

#82 - How to print the name of the objects detected in the image?

Issue - State: closed - Opened by CharlesNJ about 3 years ago - 2 comments

#81 - Predict just the classes you like

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

#80 - About multiscale inference and fusion

Issue - State: open - Opened by TonyStarkLoveYou3000 about 3 years ago

#79 - IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)

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

#78 - RutimeError: Default process group has not been initialized, please make sure to call init_process_group.

Issue - State: closed - Opened by OkuChou about 3 years ago - 2 comments

#77 - how to run this code on cpu?

Issue - State: closed - Opened by Joll123 about 3 years ago - 2 comments

#76 - Using multiscale for training and inference

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

#75 - Swin-L path and config

Issue - State: closed - Opened by MatthewCodes about 3 years ago - 4 comments

#74 - Where can I edit bounding boxes and it's outputs? Can I even change for a pre-trained model?

Issue - State: closed - Opened by CharlesNJ about 3 years ago - 5 comments

#73 - Apex is not installed

Issue - State: closed - Opened by WassimBouatay about 3 years ago - 2 comments

#72 - Couldn't get it work. I get an error when I try to run test.py

Issue - State: closed - Opened by CharlesNJ about 3 years ago - 29 comments

#71 - Are there any plans to release config files for HTC++?

Issue - State: open - Opened by snailrowen1337 about 3 years ago - 3 comments

#70 - 模型文件百度网盘的提取码

Issue - State: closed - Opened by 8125345 about 3 years ago - 2 comments

#69 - AttributeError: 'AmpState' object has no attribute 'loss_scalers'

Issue - State: open - Opened by secortot about 3 years ago - 4 comments

#68 - Download swin transformer backbone pretrained with COCO

Issue - State: closed - Opened by kyuheejo about 3 years ago - 3 comments

#67 - Inference results AP AR mAP mAR are all -1

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

#66 - 训练出现cudaErrorInvalidDevice怎么办？

Issue - State: open - Opened by dh15037672010 about 3 years ago - 5 comments

#65 - How to extract segmentation masks in the demo code?

Issue - State: open - Opened by vj697 about 3 years ago

#64 - How to change learning rate? Changing config doesn't work

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

#63 - convergence problem on cityscapes

Issue - State: closed - Opened by ngfuong about 3 years ago - 2 comments

#62 - How can I transfer training data?

Issue - State: open - Opened by admiraldus1 about 3 years ago - 2 comments

#61 - training time get slower within an epoch?

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

#60 - where are the coco pretrained checkpoints?

Issue - State: closed - Opened by lingcong-k about 3 years ago - 4 comments

#59 - KeyError: 'SwinTransformer is not in the models registry'

Issue - State: closed - Opened by MatthewCodes about 3 years ago - 5 comments

#58 - Environment for Swin-Transformer-Object-Detection

Issue - State: closed - Opened by ginobilinie over 3 years ago - 4 comments

#57 - when i run python tools/train.py configs/swin/cascade_mask_rcnn_swin_tiny_patch4_window7_mstrain_480-800_giou_4conv1f_adamw_1x_coco.py --cfg-options model.pretrained=./cascade_mask_rcnn_swin_tiny_patch4_window7_1x.pth,error appear

Issue - State: closed - Opened by Bin-ze over 3 years ago - 3 comments

#56 - I need the output channel to be a 521-dimensional vector, but I am not sure whether the number of input embedding channels of the swin network is a property that can be set by myself

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

#55 - Transformer Swin backbone to other Detector get 0 AP

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

#54 - Freeze backbone weights

Issue - State: closed - Opened by nobel861017 over 3 years ago - 2 comments

#53 - AssertionError: Default process group is not initialized

Issue - State: open - Opened by FirasKedidi over 3 years ago - 6 comments

#52 - How do you use a pre-trained model to inference a picture?Does this code support CPU?

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

#51 - Use Swin-Transformer-Object-Detection for Zero-shot object detection

Issue - State: open - Opened by snat1505027 over 3 years ago

#50 - How to achieve only object detection without instance segmentation

Issue - State: open - Opened by kekexixili over 3 years ago - 8 comments

#49 - How to extract roi features of Swin

Issue - State: open - Opened by dong03 over 3 years ago - 2 comments

#48 - AssertionError: Default process group is not initialized

Issue - State: open - Opened by FirasKedidi over 3 years ago - 8 comments

#47 - about swin pre-trained model

Issue - State: open - Opened by NEUdeep over 3 years ago - 4 comments

#46 - coco.py bug for only one class

Issue - State: open - Opened by lingcong-k over 3 years ago - 7 comments

#45 - Checkpoint Save error in google colab

Issue - State: open - Opened by Atul997 over 3 years ago - 1 comment

#44 - What is the error? There is only one class in CLASSES.I don't know if that's the reason,What should I do about it.Thanks!

Issue - State: open - Opened by kekexixili over 3 years ago - 7 comments

#43 - Onnx Conversion: Integer division of tensors using div or / is no longer supported

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

#42 - How to use COCO pre-trained weights?

Issue - State: closed - Opened by JHLEE17 over 3 years ago - 2 comments

#41 - Error during training in last iteration of first epoch

Issue - State: open - Opened by Atul997 over 3 years ago

#40 - Can you report the performance of SWIN on Retinanet under 1* and 3* schedule?

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

#39 - How to modify the number of training pictures

Issue - State: open - Opened by sure7018 over 3 years ago

#38 - mAP has some difference with the report result

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

#37 - python3.7/site-packages/mmdet-2.11.0-py3.7.egg/mmdet/apis/train.py", line 75, in <listcomp> seed=cfg.seed) for ds in dataset TypeError: object of type 'int' has no len()**

Issue - State: open - Opened by wangxiao5791509 over 3 years ago - 2 comments

#36 - Changing backbone of the model

Issue - State: open - Opened by AnhMinhTran over 3 years ago - 1 comment

#35 - KeyError: 'SwinTransformer is not in the models registry'

Issue - State: closed - Opened by Atul997 over 3 years ago - 18 comments

#34 - FLOPs and FPS measurement.

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

#33 - Mask RCNN pretrained error

Issue - State: closed - Opened by WuNein over 3 years ago - 4 comments

#32 - MoBY with Swin Transformer

Pull Request - State: closed - Opened by impiga over 3 years ago

#31 - Can I add more layers in the last stage of the backbone when I run a detection task?

Issue - State: open - Opened by 1338199 over 3 years ago

#30 - Don't save amp state if not available

Pull Request - State: closed - Opened by kentaroy47 over 3 years ago - 2 comments

#29 - IndexError list index out of range

Issue - State: closed - Opened by kentaroy47 over 3 years ago - 2 comments

#28 - Pretrained Image model in README

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

#27 - What is the hardware configuration to achieve the reported FPS?

Issue - State: open - Opened by ixez over 3 years ago - 1 comment

#26 - The model and loaded state dict do not match exactly

Issue - State: closed - Opened by vobecant over 3 years ago - 7 comments

#25 - Comment out the configuration related to the mask

Issue - State: open - Opened by sure7018 over 3 years ago - 5 comments

#24 - Gradient overflow. Skipping step, loss scaler 0 reducing loss scale to 2.926047721682624e-98

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

#23 - index out of bounds

Issue - State: open - Opened by Wangzs0228 over 3 years ago - 6 comments

#22 - missing keys in source state_dict

Issue - State: closed - Opened by jiabeiwangTJU over 3 years ago - 5 comments

#21 - The testing results of the whole dataset is empty

Issue - State: closed - Opened by sure7018 over 3 years ago - 1 comment

#20 - i have problem when using my own dataset with cascade_mask_rcnn_swin_tiny for detection

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

#19 - How to train a custom dataset?

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

#18 - The testing results of the whole dataset is empty

Issue - State: open - Opened by sure7018 over 3 years ago - 5 comments

#17 - How long does it take to train cascade Mask RCNN with Swin-B for you on COCO instances for you?

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

#16 - ERROR: Unexpected segmentation fault encountered in worker.

Issue - State: open - Opened by TfeiSong over 3 years ago - 1 comment

#15 - where is Mask RepPoints V2 config?

Issue - State: closed - Opened by mathpopo over 3 years ago - 7 comments

#14 - How to finetune the pretrained model on other COCO-format dataset?

Issue - State: open - Opened by bobo199830 over 3 years ago - 2 comments

#13 - using the config and pretrained files for detection

Issue - State: open - Opened by giorgiozannini over 3 years ago - 6 comments

#12 - SyntaxError: invalid syntax 'dataset_type': 'CocoDataset' lib2to3.pgen2.parse.ParseError: bad input: type=3, value="'dataset_type'", context=('\n', (148, 0))

Issue - State: open - Opened by leiqing1 over 3 years ago - 7 comments

#11 - TypeError: CascadeRCNN: SwinTransformer: init() got an unexpected keyword argument 'depth'

Issue - State: closed - Opened by caiduoduo12138 over 3 years ago - 1 comment

#10 - convergence problem for Cascade Mask RCNN

Issue - State: open - Opened by GrassBro over 3 years ago - 4 comments

#9 - 在构建检测器的时候显示：KeyError: "CascadeRCNN: 'SwinTransformer is not in the backbone registry'"

Issue - State: closed - Opened by TfeiSong over 3 years ago - 4 comments

#8 - Custom Dataset Training Runtime error

Issue - State: closed - Opened by HelyModi over 3 years ago - 4 comments

#7 - Model usage in an FPN-architecture

Issue - State: closed - Opened by kzhang0718 over 3 years ago - 1 comment

#6 - detect without mask

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

#5 - no trick version？

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

GitHub / SwinTransformer/Swin-Transformer-Object-Detection issues and pull requests