Questions tagged [image-processing]

A form of signal processing where the input is an image. Usually treating the digital image as a two-dimensional signal (or multidimensional). This processing may include image restoration and enhancement (in particular, pattern recognition and projection).

Filter by
Sorted by
Tagged with
0
votes
0answers
9 views

Clustering points (belonging to polygonal chains) with unknown number of clusters

I have a set of polygonal chains $\{p_1, \dots, p_n\}$ that lie in a plane. $n$ is in the range of $100-200$. Each polygonal chain has around two to five points and represents a handwritten small ...
0
votes
2answers
20 views

Connection between grey scale and intensity in image matrices

Let A be a image matrix where each element in the matrix displays the integer value of the pixel intensity for a specific pixel. We also have a greyscale between two values. For example let A be a ...
1
vote
1answer
15 views

Transfer learning on images with higher dynamic range

Is it possible to fine-tune a CNN-based model previously trained on grey-scale images with 8 bits depth [0 ~ 2^8] to fit a 16 bits depth [0 ~ 2^16] images? if there is any research paper that confirm ...
4
votes
1answer
197 views

Why does torchvision.models.resnet18 not use softmax?

I see image-classification models from torchvision package don't have a softmax layer as final layer. For instance, the following snippet easily shows that the resnet18 output doesn't have a sum = 1, ...
0
votes
0answers
16 views

Why is self-attention used for image classification?

I'm wondering why you would use Self-Attention across an entire image for image classification. What are the advantages of Self-Attention compared to a pure MLP?
0
votes
0answers
5 views

How can I calculate loss if the size of the true image is different from the output in mask rcnn?

Refer to the mask rcnn paper When the feature map passes through the roi align layer, it has a fixed 14x14. The output after the upsampling layer is increased to 28x28. In other words, the size of the ...
0
votes
0answers
15 views

Clustering into hours-of-the day, using snapshots of the environment

Let's say I have hourly measurements of 2 variables, taken from all hours of the day. I wish to classify these into (ideally) 24 bins, so that when I get a new snapshot, I might assign it to its ...
1
vote
0answers
10 views

Summary of all the results of all participants after 2015 in ImageNet (ILSVRC) challenge. Does such resource exist?

I have been reading about the ImageNet Large Scale Visual Recognition Challenge (ILSVRC). And while I can find the paper by Russakovsky et al. from 2014(updated in 2015), which contains all the ...
0
votes
0answers
16 views

spherical embedding of image data

Embedding positive data images onto a spherical manifold using the spherical embedding method doesn't guarantee the generation of positive directional quantities as the embedding positions are ...
0
votes
0answers
16 views

What is the filter experiment?

I read a method to check which model does best in measuring uncertainty. Source: page 485 of Probabilistic Deep Learning: with Python, Keras and Tensorflow Probability The steps are as follows: ...
1
vote
0answers
23 views

Edge Detection Convolution Intuition

I was learning about convolution and how filtering helps us to detect an edge in an image;however I still cannot not understand how the convolution process in the image below does this. I understand ...
1
vote
1answer
29 views

Is the density of PixelCNN normalized?

PixelCNN++ constructs a model distribution $p(x)$ over images $x\in\mathbb{R}^{n\times n}$ as a product of conditional distributions over pixels $$p(x)=p(x_1,...,x_{n^2})=\prod_{i=1}^{n^2} p(x_i| x_1,....
0
votes
0answers
10 views

Would preprocessing an image with multiple filters speed up training?

I came across this post - https://stackoverflow.com/questions/23470229/why-do-i-must-use-sobel-operator/23478399#23478399, where they use a Sobel filter for improving handwritten digit classifiers. ...
0
votes
0answers
35 views

Comparing colors of digital and print (in cloth) images

I want to compare color histograms between a digital photo and its physically printed in cloth (like a tshirt) version. The process is simple: Printing a digital image (with good resolution) in a ...
1
vote
1answer
36 views

How Convolutional layer work exaclty in RGB image processing?

I'm studying convolutional layers and I'm pretty confused. Supposing that I give to my network (CNN) an RGB image, so an image with three channels. Since the image has 3 channels, then the kernels ...
0
votes
0answers
8 views

Neural Network Architecture for Image to Image Supervised Learning?

I have a task where I would like to create a supervised learning model where each training record of X,y is a pair of images. e.g. learning a transformation from an image to an image. Is there a ...
1
vote
0answers
31 views

Validation accuracy drops with ResNet50 augmented training

I am trying to train a ResNet50 from scratch on Imagenette https://github.com/fastai/imagenette I started by just directly training and had the result below: Orange: training dataset. Blue: ...
2
votes
1answer
47 views

Contour plot comparison [closed]

What are the ways to compare two different maps? I was thinking of taking the matrices of each map and then quantify their spatial differences by plotting the map of the differences of the two ...
0
votes
0answers
35 views

High test accuracy but low performance on new samples

am working on an image classification application and get good results about 97% test accuracy but when I test the trained model on a new unseen sample the model failed to predict the right class ...
1
vote
0answers
28 views

How to determine the rotation of an objects relative to another?

I have two images. One with a rotated object, one with a non-rotated object. How can I measure the degree of rotation of the second object with respect to the first using Python? I tried doing ...
0
votes
0answers
5 views

How to give affinity propagation a sense of context

I have a task where affinity propagation helps me cluster similar looking visual regions in an image. Say I have a fruit bowl with apples and bananas and I have some way of getting bounding boxes ...
0
votes
0answers
10 views

Calculate the weight of a WLS solver and evaluate an Image-To-Image Training

I have a depth map reconstruction problem that I have formulated in this form : With w(u,v) is the weighting term. To produce the two maps R and W I used a convolutional neural network. Now, I ...
0
votes
0answers
41 views

Learning Architecture for Very Large Image Outputs?

I have a large number of training examples $(x_i,y_i)$ where $x_i \in \mathbb{R}^{n \times n}$ and $y_i \in \mathbb{R}^{n^2 \times k}$, where $n$ is a number like $512$ and $k$ is a smaller number ...
1
vote
0answers
23 views

SSIM loss with possibly negative input

I have a regression task (image reconstruction) and the benchmark was done using L1Loss (Pytorch). The input images were normalised to [-1, 1] so that tanh activation could be used. When I tried ...
0
votes
0answers
36 views

the accuracy grid search gave me for image classification (using feature extraction with vgg16 and xgboost) was wrong?

so I'm somewhat a beginner at machine learning. for an image classification problem, I used feature extraction using vgg16 and gave the features to xgboost model as input. then used grid search to get ...
2
votes
0answers
18 views

How to extract simple shapes from a feature map?

I am working on image parsing project. I want to find a way to automatically parse an object into a list or a graph of simpler shapes. Is there any practical information on how to do so? So far I took ...
1
vote
0answers
60 views

L1 loss giving a better result than L2 loss for optimizing PSNR in an image super resolution problem

in an Image Super Resolution kind of problem, I want to get the highest PSNR values for the super resolution images from the low resolution images obtained after training a model. I experimented with ...
0
votes
0answers
12 views

How to compute a spatial covariance matrix within a cluster?

I'm trying to implement this paper, which is a method to obtain superpixels through a SLIC-based approach, and at some point I need to calculate a spatial covariance matrix for each cluster - or ...
0
votes
0answers
5 views

How to combine two different sets of attributes in affinity propagation

I'm using affinity propagation to cluster visually similar AND spatially close regions in an image. I already have a detection algorithm which places bounding boxes around regions of interest. From ...
0
votes
0answers
17 views

Calculate the coherence with matplotlib.mlab.cohere

My overall aim is to compare the edges of two images by comparing their Fourier Transforms (FFT) and to calculate one number as a key performance indicator that describes how much they are similar to ...
0
votes
0answers
36 views

Relation between Fourier transforms and coherence of signals

My overall aim is to compare the edges of two images by comparing their Fourier Transforms (FFT) and to calculate one number as a key performance indicator that describes how much they are similar to ...
0
votes
0answers
10 views

Cross-validation for training an SVM classifier on fisher vectors

I have an image classification problem with three classes, each with about 650 training images. Currently, I am using various feature extractors (SURF, HOG, LBP...) to extract features from the images....
1
vote
0answers
15 views

What is the influence of the number of channels on the input of a CNN? [closed]

Let's say I want to pass an RGB image through a CNN. It has three channels, each containing a 8 bit integer per pixel, ranging from 0 to 255. Let's say I know encode this three channels in only one ...
0
votes
0answers
15 views

How to augment 3D medical images for pytorch model

I have medical images in dicom format. I have read the images and extracted the pixel values and stored as a numpy array. The final shape of array of a single patient is in the form [channels, frames/...
0
votes
0answers
17 views

Poor accuarcy score for Semi-Supervised Support Vector machine

I am using a Semi-Supervised approach for Support Vector Machine in Python for the image classification from PASCAL VOC 2007 data. I have tried with the default parameters from the libraries and also ...
3
votes
0answers
182 views

What is p(data) in image generation

In the context of image generation architectures such as VAEs or GANs (say we are using mnist digits) what do we mean by probability distribution of the data? Just to clarify this question and make it ...
0
votes
0answers
20 views

Implicit Neural Representations doubts

In the paper Implicit Neural Representations with Periodic Activation Functions it is explained that the aim of Implicit Neural Representations is to learn the parameters of a neural network such that ...
1
vote
1answer
21 views

What happens to the weights of a pretrained model on transfer learning?

I want to use a pretrained neural network for two similar (but not identical) classification problems. Let's say I want to use AlexNet for image classification, where in problem A I am interested in ...
0
votes
1answer
21 views

How to derive the high dimensional image from low dimensional one without having original data (image)?

Hi! I am new to Computer Vision. I am curious as to whether it is possible to decompress a low dimensional image to a high dimensional one if I don't have an original image. What methods are ...
0
votes
0answers
15 views

Deciding cutoff threshold for background in image recognition

I have data in the form of intensities for each cell in a fluorescent staining project, and I am wondering how I can filter real signal apart from background noise. There is a software that I used to ...
0
votes
0answers
37 views

What image filter is used in max pooling in image processing

I am curious to know which filter is used in to do max pooling? I am aware that it is a Deconvolution layer. As it takes the maximum value across all pixels - would it use an nonlinear area filter? As ...
0
votes
0answers
32 views

Correct way of computing dice score for image segmentation?

In binary image segmentation, for given a set of images, it's true mask and predicted mask. How to compute dice score?, should I compute dice score for each image separately and then find mean across ...
0
votes
0answers
25 views

Autoencoder failed to find anomaaly

We're using the following architecture for anomaly-finding- ...
2
votes
0answers
29 views

What could be a good metric for rotation/translation-invariant image similarities?

An acquaintance posted an interesting image processing question. In the image below, we can tell that (a) and (c) are similar, whereas (b) is different. If we use (a) as a template and want to have a ...
0
votes
1answer
42 views

Which statistical test when you have two reviewers (for qualitative scores)?

Two reviewers (radiologists) assessed image performance (resolution, signal/noise, contrast...) based on a 5-point scale : 1, markedly worse; 2 worse; 3 equivalent; 4 better; markedly better) of 23 ...
0
votes
0answers
20 views

Is validation performance sufficient for hyperparameters choice on a small dataset for images multi-classificaiton problem

Problem: multiclass (3-6 classes) images classification (DeepLearning). Dataset size <2000 samples. One class is rare <50 samples. We've conducted several sets of stratified cross-validations ...
0
votes
1answer
12 views

"Conorm" image preprocessing

For a personal project, I am trying to reproduce the MCDNN of the paper Multi-column Deep Neural Networks for Image Classification by Ciresan, Meier, and Schmidhuber on CIFAR-10. The different ...
0
votes
0answers
33 views

Validation and training dataset partition of video frames

I'm training a deep neural network to colorize grayscale images. The dataset consists of frames extracted from a video in order; for every frame there's a grayscale image (input) and a colored image (...
1
vote
0answers
21 views

Bilinear Interpolation Algorithm for up-sampling 2D images

In keras it is possible to use UpSampling2D layer to up-sample an image. You can use Bilinear Interpolation. Given an image ${h\times w}$ it is possible to increase its size in ${h*k\times w*l}$, ...
0
votes
0answers
8 views

Patch wise feature vector comparison

I have a image of size of 64*64. I am trying to compute HOG features for the image. I have skimage for my implementation, with the following parameters: ...

1
2 3 4 5
12