Questions tagged [semantic-segmentation]
The semantic-segmentation tag has no summary.
50 questions
0 votes
0 answers
15 views
UNETR paper — Does brain tumor segmentation use 5-fold CV or 80:15:5 split?
I'm reading the UNETR paper ("UNETR: Transformers for 3D Medical Image Segmentation") and I'm confused about the training and evaluation methodology for the brain tumor segmentation task on ...
2 votes
0 answers
28 views
Clarification on “random scale” in UNETR data augmentation
I am trying to replicate the paper "UNETR: Transformers for 3D Medical Image Segmentation". In the Implementation Details section, they wrote: In addition, we used data augmentation ...
8 votes
1 answer
160 views
How to correctly implement the loss function for my distillation of Mask2Former?
I have a Mask2Former model fine-tuned on my own custom dataset and it is working nicely. I want to play around with knowledge distillation and use my pretrained ...
0 votes
0 answers
19 views
decreased prediction accuracy of segmentation model on using smaller size but higher resolution images
I am using a the segmentation models library and used a UNET model with efficientnet-b7 backbone pretrained on imagenet, to segment immunohistochemistry images.I used augmentations to increase the ...
0 votes
0 answers
34 views
Am I following the correct approach?
I am working on a medical semantic segmentation problem, where the input is of size (1,256,256) and I have two outputs associated with that input ( two masks ) each with shape ( 1,256,256). For every ...
1 vote
0 answers
56 views
Per Channel loss or Per Sample Loss
I am currently tackling a semantic segmentation problem where, for each sample, my goal is to segment two masks corresponding to two objects. Notably, object two is typically located inside object one,...
1 vote
0 answers
56 views
How do transposed convolutions in CNNs reduce the channel dimensionality?
In CNNs, I understand how convolution works and how it gradually reduces spatial resolution but increases the channel dimension. E.g. an RGB image of 100x100x3 after a few convolution layers may ...
0 votes
0 answers
70 views
Why can't I increase my GPU utilization?
I have a simple UNet model (~1M params) written in Keras 3.0.1, running with a torch backend. My CUDA version is ...
0 votes
1 answer
123 views
How do I ensure final output shape matches input shape for a semantic segmentation task?
I trying to replicate the semantic segmentation example https://keras.io/examples/vision/oxford_pets_image_segmentation/ but train on my own data. I have 8 labels (7 features + background). My images ...
0 votes
0 answers
65 views
Trying to train a denoising autoencoder to restore missing information from a binary image
I am building a denoising autoencoder to repaint lanes from a binary image. The input is a binary image that has incomplete lanes, due to vehicles getting in the way. I repaint the lanes manually so ...
0 votes
0 answers
180 views
Semantic segmentation sometimes give bad result
I'm training Unet+MobileNetV3 for semantic segmentation objects on real photos using custom dataset and get strange results. I have already accumulated pretty big dataset and constantly improve it by ...
1 vote
1 answer
84 views
Building a CNN (with Keras for pixelwise classification)
I have a set of 120x120 input images with 3 channels. I want to build a basic CNN to predict the value of each pixel. I have 2 doubts. One is regarding the last layer - should be a Dense layer, or a ...
1 vote
2 answers
1k views
Why not using segmentation architectures for object detection?
Current object detection architectures like Faster-RCNN and YOLO seem to be overcomplicated in comparsion with segmentation architectures like Unet. So, why can't we just draw some rectangles around ...
0 votes
1 answer
68 views
MaskRCNN.train gives 'list index out of range'
I have been trying to use MaskRCNN with a Resnet backbone on the DeepFashion2 Dataset for instance segmentation. The custom configurations are as follows: ...
1 vote
0 answers
362 views
Meta's SAM model: can extract semantic embedding vector?
I'm interested in finding an embedding vector for each segment found by the Facebook/Meta Segment Anything model (for classification and tracking of segments). Can ...