190 questions
0
votes
0
answers
33
views
UNet is loading 1 epoch per day [closed]
I am not allowed to share my full code but I can share parts.
I have built a UNet that uses cache and shuffle buffers. It was working well and training fast until I realised any updates I made were ...
0
votes
0
answers
25
views
Checkerboard Mosaic When Merging Prediction Tiles from TensorFlow UNet
I’m running a TensorFlow UNet model on geospatial imagery. I split the image into overlapping 500×500 pixel tiles using a custom function (get_tiles) and use a stride of 460 (i.e. 500 – 2×padding) so ...
- 1
0
votes
0
answers
12
views
U-Net model: incorrect number of inputs and outputs, find error in class code / architecture
I am trying to get a custom version of the EelSpecNet model working, which is based on the Unet architecture. The original large version can be found here:
text
The model ist used to reconstruct an ...
- 21
1
vote
1
answer
51
views
Unet pytorch dimension mismatch
I got the following U-net architecture causing problems:
class UNet(nn.Module):
def __init__(self, in_channels, out_channels):
super(UNet, self).__init__()
self.encoder1 = self....
- 35
2
votes
1
answer
60
views
MethodError in Training Minimal U-Net
I need to implement a U-Net in Julia using Flux.jl. The ultimate goal is to train a neural network for a scientific problem. As a first step, I decided to experiment with the KITTI benchmark dataset.
...
- 23
0
votes
0
answers
73
views
Differences in no of ResNet blocks in up blocks and no of channels for Unet2D model of diffusers
I have been reading about Unets and Stable diffusion and want to train one. I understand the original architecture for unets and how its channels, height and width evolve over down blocks and up ...
- 21
0
votes
0
answers
46
views
How do I use "concatenate" in a Keras model without getting shape related error?
I'm trying to create a U-Net architecture in Keras 2.15. The issue arises when I introduce the "concatenate" layers. The error seems to be related to the tensor shape being passed through ...
- 1
0
votes
0
answers
32
views
Unet discrepency between input and output shape
I am working on a U-Net model with two outputs: segmentation and classification. My dataset has 110 object classes in coco format, and I'm using TensorFlow/Keras to train the model. The segmentation ...
0
votes
0
answers
23
views
I want to put the calculated voxel into my csv file but i can't put it in just as the code in dl+direct
i am currently working on a project, and i want to use the model of swin unetr and the way to save the output just as dl+direct. the purpose is to do brain segmantations
however when i put different ...
- 1
0
votes
1
answer
62
views
Difficulty fixing the code for DDPM with unet using Pyorch
I am studying Deep learning and have an assignment for training a DDPM with UNet on the MNIST handwritten digits dataset. Three ipynb files (model, unet and train_mnist) and one pdf for unet diagram ...
- 1
0
votes
0
answers
17
views
Approaches to reduce splotchiness on UNet output
I'm training a UNet to try and predict caustics, kind of as a toy problem.
An input image to the UNet is something like this:
It is trained on "ground truth" outputs like this:
It is ...
- 341
0
votes
0
answers
84
views
Executing a 3D U-net I get widely different metrics in each execution, sometimes >99,99% across accuracy, recall, DICE and IoU
I have created a custom 3D U-Net for multi-class semantic segmentation. Currently, I'm using it for medical image classification, where only two classes are present. I have two related issues: with ...
0
votes
1
answer
93
views
ValueError: Invalid input shape when combining image, mask, and CSV data in Keras model
I’m working on a deep learning model using Keras where I’m trying to combine three inputs: images, masks, and CSV data. The goal of my model is to predict the presence and type of brain hemorrhage ...
0
votes
1
answer
53
views
Add out_channel to pretrained model
I have trained a model fish.pth with 2 channels (fish and background). I want to add a layer so that I train a new model spine.pth which segments 3 classes fish, spine and background. Yes, I can do it ...
- 1
0
votes
0
answers
143
views
Residual Connection in Pytorch
I have a VNET network (see here for reference) There are two types of skip connections in the paper. Concatenating two tensors and element wise add. I think i am implementing the second one wrong, ...
- 303