milesial/Pytorch-UNet issues and pull requests

#297 - Has anyone tried implementing raytune for hyperparameter optimization?

Issue - State: open - Opened by k-nayak about 3 years ago - 12 comments

#270 - How to output validation args such as miou, f1_score, accuracy, recall and precision?

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

#258 - Assertion `t >= 0 && t < n_classes` failed RuntimeError: CUDA error: device-side assert triggered

Issue - State: closed - Opened by pustar almost 4 years ago - 3 comments

#242 - Help with training on CARVANA images and masks

Issue - State: closed - Opened by rtrahms almost 4 years ago - 9 comments

#198 - TypeError: to() received an invalid combination of arguments

Issue - State: closed - Opened by wjj001-dl over 4 years ago - 3 comments

#173 - Dice coefficient no change during training,is always very close to 0

Issue - State: open - Opened by rameses666 over 4 years ago - 29 comments

#101 - Tested masked are all zeros.

Issue - State: closed - Opened by August-us almost 5 years ago - 5 comments

#100 - dice coefficent very small

Issue - State: closed - Opened by cmlyldz almost 5 years ago - 1 comment

#99 - dice coeff or loss function gathering

Issue - State: closed - Opened by cmlyldz almost 5 years ago - 1 comment

#98 - TypeError: 'int' object is not iterable when running code predict,py

Issue - State: closed - Opened by BianChang almost 5 years ago - 7 comments

#97 - Can you provide the model weight?

Issue - State: closed - Opened by Ttt749 almost 5 years ago - 1 comment

#96 - Problem with starting train

Issue - State: closed - Opened by cmlyldz almost 5 years ago - 4 comments

#95 - predict result is black

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

#94 - Some problems occurred in predict.py.

Issue - State: closed - Opened by DL-Alva almost 5 years ago - 21 comments

#93 - RuntimeError: Expected object of scalar type Long but got scalar type Float for argument #2 'target'

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

#92 - Batch loss results

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

#91 - i have some problems in how to evaluate segmentation accuracy in eval.py

Issue - State: closed - Opened by MHX1203 almost 5 years ago - 1 comment

#90 - Unet

Issue - State: closed - Opened by Zebrocode almost 5 years ago - 2 comments

#89 - what version of numpy is used,and what should be filled in this place？？？

Issue - State: closed - Opened by duanliangtao almost 5 years ago - 1 comment

#87 - my datasets have 6 classes , i find that the network has so poor performance

Issue - State: closed - Opened by XUYUNYUN666 almost 5 years ago - 3 comments

#86 - train.py RuntimeError: expected input[1, 1, 307, 409] to have 3 channels, but got 1 channels instead

Issue - State: closed - Opened by alkeksibusiii almost 5 years ago - 2 comments

#85 - imgs = np.array([i[0] for i in b]).astype(np.float32) ValueError: could not broadcast input array from shape (3,288,404) into shape (3)

Issue - State: closed - Opened by XUYUNYUN666 almost 5 years ago - 5 comments

#84 - loss is negative

Issue - State: closed - Opened by fengjiejiejiejie almost 5 years ago - 2 comments

#83 - Hi,milesial!

Issue - State: closed - Opened by fengjiejiejiejie almost 5 years ago

#82 - can u provide the pre-trained modles in your research?

Issue - State: closed - Opened by hanyeliu almost 5 years ago - 6 comments

#81 - Help with results

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

#80 - Default parameters to run the program，but the output picture is chaotic!!!

Issue - State: closed - Opened by duanliangtao about 5 years ago - 4 comments

#79 - UserWarning: Using a target size (torch.Size([1, 3, 224, 224])) that is different to the input size (torch.Size([1, 7, 224, 224])) is deprecated. Please ensure they have the same size.

Issue - State: closed - Opened by bilibili2007 about 5 years ago - 3 comments

#78 - When i run this code ,get loss is negative??

Issue - State: closed - Opened by YoudianguaiYi about 5 years ago - 4 comments

#77 - F.interpolate is a function not a class

Issue - State: closed - Opened by CXMANDTXW about 5 years ago - 2 comments

#76 - ValueError: Target and input must have the same number of elements. target nelement (1373880) != input nelement (457960)

Issue - State: closed - Opened by huskyachao about 5 years ago - 7 comments

#75 - nn.functional.sigmoid is deprecated. Use torch.sigmoid instead.

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

#74 - Activation function

Issue - State: closed - Opened by wujiayi about 5 years ago - 2 comments

#73 - The question about the function "get_square" in “utils.py”

Issue - State: closed - Opened by scmales about 5 years ago - 2 comments

#72 - Error in merge_masks

Issue - State: closed - Opened by mira-murali about 5 years ago - 5 comments

#71 - Grayscale images unexpected weight dimensions error is hard to traceback

Issue - State: closed - Opened by t2ac32 about 5 years ago - 2 comments

#70 - Why the predicted image's background is black?

Issue - State: closed - Opened by W998K about 5 years ago - 3 comments

#69 - unet_model in the code is a little different form the picture of the Unet

Issue - State: closed - Opened by xiaochengfuhuo about 5 years ago - 6 comments

#68 - Why use padding=1 not 0?

Issue - State: closed - Opened by shoujieqiu over 5 years ago - 2 comments

#67 - over-fitting

Issue - State: closed - Opened by AbnerAI over 5 years ago - 1 comment

#66 - TypeError: unsupported operand type(s) for /: 'tuple' and 'int'

Issue - State: closed - Opened by AbnerAI over 5 years ago - 1 comment

#65 - Negative Loss

Issue - State: closed - Opened by VoiceBeer over 5 years ago - 2 comments

#64 - what is the performance of the given model trained on your dataset?

Issue - State: closed - Opened by zst1992 over 5 years ago - 1 comment

#63 - could not broadcast input array from shape (3,384,384) into shape (3)

Issue - State: closed - Opened by AbnerAI over 5 years ago - 4 comments

#62 - Function Error

Issue - State: closed - Opened by WxpHB over 5 years ago - 2 comments

#61 - CUDA Out of memory

Issue - State: closed - Opened by Arktius over 5 years ago - 8 comments

#60 - Why not use the PyTorch DataLoader?

Issue - State: closed - Opened by Dr-Cube over 5 years ago - 4 comments

#59 - Your network seems to be different from the paper

Issue - State: closed - Opened by ardeal over 5 years ago

#58 - Target and input must have the same number of elements

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

#57 - Does it work with greyscale training images?

Issue - State: closed - Opened by mgersalan over 5 years ago - 2 comments

#56 - Dice computation over 1

Issue - State: closed - Opened by Mihai196 over 5 years ago - 4 comments

#55 - Is it your idea to add batch normalization and replace upconv with bilinear upsample?

Issue - State: closed - Opened by fanfanda over 5 years ago - 4 comments

#54 - custom dataset

Issue - State: closed - Opened by vcvishal over 5 years ago - 10 comments

#53 - why I got bad result

Issue - State: closed - Opened by hac135 over 5 years ago - 4 comments

#52 - Problem about GPU memory

Issue - State: closed - Opened by Ruikun-Li over 5 years ago - 5 comments

#51 - Masks format

Issue - State: closed - Opened by mehditlili over 5 years ago - 4 comments

#50 - Model.pth

Issue - State: closed - Opened by vcvishal over 5 years ago - 2 comments

#49 - why the out of self.up1 is 256 but 512 , and the out of self.down4 is 512 but 1024?

Issue - State: closed - Opened by wky1998 over 5 years ago - 1 comment

#48 - dice computation error

Issue - State: closed - Opened by andyaloha over 5 years ago - 2 comments

#47 - ValueError: axes don't match array

Issue - State: closed - Opened by I-CANT-CODE over 5 years ago - 9 comments

#46 - fix split_ids (returned only (id, 0))

Pull Request - State: closed - Opened by pierrezeb almost 6 years ago - 2 comments

#45 - index + 1 in average

Pull Request - State: closed - Opened by dmitrysarov almost 6 years ago - 1 comment

#44 - batch size mismatch or spatial target dimensions

Issue - State: closed - Opened by CodLiver almost 6 years ago - 1 comment

#43 - How to initialize the parameters of the Unet

Issue - State: closed - Opened by mlx123 almost 6 years ago - 1 comment

#42 - What the learning rate should be?

Issue - State: closed - Opened by ChaoLi977 almost 6 years ago - 2 comments

#41 - batch dimension in eval.py

Issue - State: closed - Opened by dmitrysarov almost 6 years ago - 4 comments

#40 - Out of memory

Issue - State: closed - Opened by ChaoLi977 almost 6 years ago - 7 comments

#39 - How's the performance of this implementation

Issue - State: closed - Opened by eglxiang almost 6 years ago - 4 comments

#38 - upsample

Issue - State: closed - Opened by zhangsdly almost 6 years ago - 1 comment

#37 - Move the sigmoid activation to the model itself

Pull Request - State: closed - Opened by rht almost 6 years ago - 1 comment

#36 - when i exec python predict.py -i test.jpeg -o output.jpeg report erro

Issue - State: closed - Opened by daixiangzi about 6 years ago - 1 comment

#35 - error about backward in dice_loss.py

Issue - State: closed - Opened by bobo0810 about 6 years ago - 2 comments

#34 - How to prepare the training data?

Issue - State: closed - Opened by wangzhe0623 about 6 years ago - 4 comments

#33 - Pushing version 2 of modifications

Pull Request - State: closed - Opened by mamemo about 6 years ago

#32 - Can we change the U-net to fit any size of the input image?

Issue - State: closed - Opened by j96w about 6 years ago - 2 comments

#31 - Switch net.train() to net.eval() during evaluation?

Issue - State: closed - Opened by j96w about 6 years ago - 2 comments

#30 - Network Precision and recall.

Issue - State: closed - Opened by allenhamm about 6 years ago - 1 comment

#29 - Trying to use it with other dataset.

Issue - State: closed - Opened by bermudezarii about 6 years ago - 6 comments

#28 - Thanks for the great repository!

Issue - State: closed - Opened by chaoyuaw about 6 years ago - 1 comment

#27 - the return value maybe wrong? in load.py

Issue - State: closed - Opened by yelusaleng about 6 years ago - 7 comments

#26 - could not broadcast input array from shape (4,260,260) into shape (4)

Issue - State: closed - Opened by LXYTSOS about 6 years ago - 3 comments

#25 - prediction error

Issue - State: closed - Opened by lukui1995 about 6 years ago - 4 comments

#24 - Error in dice loss computation

Issue - State: closed - Opened by ghost about 6 years ago - 1 comment

#23 - Hello, thanks for your code!!! But I have some trouble using it

Issue - State: closed - Opened by TonyDongLB over 6 years ago - 2 comments

#22 - Dense CRF problem

Issue - State: closed - Opened by gljhd over 6 years ago - 1 comment

#21 - Incompatibility with Python2.7

Issue - State: closed - Opened by vtddggg over 6 years ago - 3 comments

#20 - Fixed argument actions

Pull Request - State: closed - Opened by changjo over 6 years ago - 1 comment

#19 - Hello how can I train your model with just one channel image(black and white)

Issue - State: closed - Opened by SWKoreaBME over 6 years ago - 6 comments

#18 - Results from a trained model of mine.

Issue - State: closed - Opened by changjo over 6 years ago - 8 comments

#17 - Problem occurs when I run predict.py

Issue - State: closed - Opened by wyzjack over 6 years ago - 4 comments

#16 - Multi class branch

Pull Request - State: closed - Opened by wangxinyu199306 over 6 years ago - 3 comments

#15 - Multi-class issues and discussion

Issue - State: closed - Opened by wangxinyu199306 over 6 years ago - 72 comments

GitHub / milesial/Pytorch-UNet issues and pull requests