Questions tagged [computer-vision]

Questions related to image representation, segmentation, visual object categorization and image processing algorithms in general.

Filter by
Sorted by
Tagged with
1
vote
2answers
256 views

In CNN, how to map from fully connected layer to output image?

In CNN, where the last layers are fully connected, how to make pixel-wise prediction to output an image(binary matrix), if the number of neurons in the last layer is less than the size of the image?
1
vote
1answer
1k views

Deep learning models for unsupervised semantic segmentation

I am working on semantic segmentation for satellite images using keras and python. It is my understanding that popular models like U-Net require mask images (labels). Are there any unsupervised deep ...
1
vote
1answer
250 views

How can we relate the concepts of GAN/cGAN in SRGAN? Is SRGAN a Conditional GAN?

I have been reading and looking at implementations of the SRGAN, from "Photo-realistic Single Image Super Resolution with Generative Adversarial Networks" paper. One thing that I noticed is that the ...
1
vote
1answer
52 views

Differentiable method to convert voxel representation into pointclouds?

I'm finding a way to convert 3D voxel data into 3D point-clouds. Since I'd like to do it inside a deep-learning architecture, the conversion has to be differentiable. Is there such a method? Thanks ...
1
vote
1answer
24 views

Visualizing a multilayered perceoptron for image classification

As part of a project I'm working on, I have built a "fully connected layer" (multilayered perceptron) network for image classification. Even though I know how to build an convolutional NN, for various ...
1
vote
1answer
100 views

Non overlapping Clustering/ Segmentation of image points

I have this simple image from map building that I wish to cluster to extract the black dots as points in image co-ordinates (ie. the x,y coordinates of the cluster centers). I have tried many ...
1
vote
2answers
406 views

After training a classifier on N classes, how do you add an N+1th class?

Say we have data for N classes and train a classifier. Then we have a new set of data for a N+1th class. How do I train a classifier that now predicts all N+1 classes? Setting is in object detection, ...
1
vote
1answer
62 views

Is it possible to find the similarity in images by only based on available colour intensity in the images ?

Is it possible to find the similarity in images by only based on available colour intensity in the images ? What are the challenges regarding this problem ? Is there any way to find a solution for ...
1
vote
1answer
157 views

What is the meaning of a P100 score in image recognition / categorization?

I've seen it used in a couple papers, none of which are freely available, but a preview gives a couple sentences involving a P100 score here: https://scholar.google.com/scholar?q=p100+score&btnG=&...
1
vote
1answer
1k views

How does faster RCNN RPN propose region on the feature maps?

Faster RCNN consists of two modules. (a) Region proposal network and (b) Fast R-CNN detector The paper mention Region proposal network runs on the feature maps. So, I have a input 256 x 256 image ...
1
vote
1answer
53 views

Reference Request - Machine Learning for Grouping Polygon Vertices with Occlusion

I will present a stripped-down version of a pattern recognition problem that I wish to solve. I would like references to machine learning algorithms and/or problem representations to address this kind ...
1
vote
1answer
490 views

Does it make sense to use GIST features as input for ConvNet?

I've been given a set of labeled images whose description arrays are 1x960 dimensional. Does it make sense to use these as input for my cnn? the 1x960 array is an output of the GIST feature extraction....
1
vote
2answers
790 views

Can we use Bag of Visual Words to compute similarity between images directly?

I'm implementing a Content Based Image Retrieval application (CBIR). I've read about the Bag of Features model and it's considered an intermediate-step algorithm in some application. For example, ...
1
vote
2answers
95 views

What advices do you have for a starter in multiple image recognition?

So, I have experience in machine learning for NLP and a little in neural networks for NLP, but never so far done anything in computer vision in this area so bear with me if what I am asking is a ...
1
vote
1answer
351 views

Restricted Boltzmann Machine for grayscale images

I know that RBM's have been used on image data for pre-training neural nets, but all I can find are RBM's on black and white images. How do you apply them to say 256 grayscale?
1
vote
0answers
18 views

What happens when the filter size is same as that of image size in a CNN?

In Convolutional Neural Networks (CNNs), theoretically speaking, if the filter dimensions are same as that of the image (training example) dimensions, will CNNs boil down to normal Fully Connected ...
1
vote
0answers
11 views

How to visualize 3d joints of a SMPL model based on pose params

I am trying to use demo.py in https://github.com/nkolot/GraphCMR. I am interested in obtaining joints from the inferred SMPL image and visualize it similar to ...
1
vote
0answers
19 views

Pre-training a network on significant landmarks

Let's say there is a dataset consisting of photos of cars from various brands, and we're trying to train a ConvNet to identify the brand from a photo, just like these ones: One approach I was ...
1
vote
2answers
17 views

How can a given conv neural net layer handle filters of different size?

traditional method is to use multiple filters of same dimensions but with different weights and stack the output (basically concatenate them) that is then to be fed into the next conv layer. If I ...
1
vote
1answer
21 views

Looseness of the definition of domain adaption

I am a bit intrigued about “domain shift” concept. Specifically, in part 5 of the paper “Coupled Generative Adversarial Networks”, it reads We studied the problem of adapting a digit classifier ...
1
vote
0answers
69 views

Karush-Kuhn-Tucker Conditions in a Biometrics Research Paper

For one of my course works I'm supposed to find a scientific paper from my field of research that involves an optimisation problem solution based on the Karush-Kuhn-Tucker conditions. My research ...
1
vote
0answers
20 views

Reconstruction using low-light images

Let's say we have a regular photo and three low-light photos illuminated in different colors. Each pixel is a three-component vector $q=(R,G,B)$. Then $q_k^{A}$ is the $k$-th pixel of the regular ...
1
vote
0answers
17 views

Understanding the concept behind Microsoft Computer Vision- Analyze an Image. Also extract metadata of image from different search engines

My task is to upload an image and get as much information as I can get about the image or in different words its metadata (not exif) like - what is it, caption, price, what does it do, application, ...
1
vote
0answers
21 views

Shape classification by color - computer vision

I have task of classification of polygons. These shapes may differ only by color, or be transformed by linear transformation (scaled, rotated). I am using ORB for separating different polygons, but it ...
1
vote
1answer
29 views

Does the model architecture of a CNN depend on the dimension of your input images?

By model architecture, I'm interested in knowing the following: Number of nodes in input layer Number of nodes in subsequent layers Number of layers in the architecture Number of filters and ...
1
vote
0answers
11 views

Finding objects in an entanglement

I'd like to know if there are ways to solve entanglement problems like this one: Given a known shape (in red), can a machine locate (or at least count) the instances of this same object in an image ...
1
vote
1answer
15 views

Same input size for Source and Target Model?

For the transfer learning do we need to have same input image size of target model as source model? for example my source model is trained on 100x100 input images and target model is low resolution ...
1
vote
1answer
404 views

How best to combine object detection and tracking

I am trying to make a computer vision system which will be able to detect and track objects of interest. This will require (1) detection functionality to notice the object when it appears (2) ...
1
vote
0answers
32 views

Image generation based on sketch

Are there any instances of image generation models, where an image (a very rough sketch) has been used as an input and was then augmented. For example: This could be a rough sketch, which is then ...
1
vote
0answers
176 views

Translation invariance of features in convolutional encoder-decoders

A big part of using convolutional layers is translation invariance, i.e., features are detected regardless of their position in the image. In a convolutional encoder-decoder that maps from an image ...
1
vote
0answers
2k views

What is the difference between dice loss vs jaccard loss in semantic segmentation task?

What is the difference between dice loss vs jaccard loss in semantic segmentation task? Dice loss: ...
1
vote
0answers
10 views

Scene annotation

I am looking to run my convolutional net on a tennis video and see if it can recognise the type of shot played( straight, volley or cross court). What tool can I use to annote the scenes in my video?
1
vote
0answers
13 views

Counting number of items in a box from a MP4 video using computer vision and R

I have an MP4 video in which a person keeps putting some household items into the box(container type). Items consist chips packets, colddrink bottles, mugs etc. Sometimes even person takes out some ...
1
vote
0answers
20 views

Segmentation Visualisation DNN

I was interested in what the network is exactly learning during the process of image segmentation. Take unet for example. In image classification, the hierarchy of features is more clearer to ...
1
vote
0answers
33 views

learning a distribution from images with a cnn

I have images which contains maybe 1000 circles of different sizes. Circles may overlap: it's not a simple picture. I'd like to do one of two things: either (1) learn the circle size probability ...
1
vote
0answers
233 views

Finding objects in an image given a single instance of the object from the same image - Deep Learning approach

I have an aerial image containing planes like this: I want to detect planes in the image (find bounding boxes around planes). For this user draws a bounding box on one of the planes and the algorithm ...
1
vote
0answers
138 views

Traffic density estimation using deep learning

I am working on a project implementing deep learning and computer vision to estimate the traffic density of any random given road segment/roundabout or intersection. I am given a camera mounted on the ...
1
vote
0answers
10 views

Classify a specific object amongst other diverse objects

I have a device which takes one picture per day of a slab. It contains many instances of a specific object (let's call it "Object A") and a few other objects (let's call them "Others"). I want to ...
1
vote
0answers
52 views

why the kernel size become greater as the spatial size of feature map goes down in inception network?

In the inception networks like inception-v3 and inception-v4, the kernel sizes are smaller in the lower layers,such as 3*3, but in the higher layers, the kernel sizes seem to be larger,such as 5*5,7*7,...
1
vote
0answers
52 views

Weight Initialization

For Neural Networks with ReLU activations it is common practice to use the initalization introduced by He et al. to keep the variance of layer activations constant throughout the network. On the ...
1
vote
0answers
226 views

What is the differennce between invariance to translation, covariance to translation and equivariance to translation?

I get stuck at understanding the difference between invariance to translation, covariance to translation and equivariance to translation in the context of of convolutional neural network. What does ...
1
vote
0answers
15 views

Performance required for histology data?

The question is as follows: How accurate does computer vision be for it to be practically useful for analyzing histology images? Reason I'm asking because I have seen hundreds of papers like ...
1
vote
0answers
84 views

Adapting a Crowd counting network for annotating objects

Good evening I have a deep learning related question, I'd be glad if somebody could share some ideas about my experiment :) Context I have created my own dataset consisting of 3D scenes containing ...
1
vote
0answers
15 views

Extracting pedestrian step duration from video

I have a front view pedestrian surveillance video of about 15 seconds. The video looks a bit like the picture bellow. I want to estimate the number of frames needed by a pedestrian to complete one ...
1
vote
0answers
32 views

Transfer learning for image classification

When working with transfer learning for image classification, I would like to freeze only a part of the convolutional base of a pretrained model while adding a classifier (some shallow network) on top ...
1
vote
0answers
93 views

How to make normalize image classification output and improve the model?

I am trying to build an image classification using transfer learning of VGG16 model. I acquired very small data set of 200 images for each class and used 10 images as validation(I know the data set is ...
1
vote
0answers
89 views

DCGAN celebA strange results

I am trying to generate attributed faces using DCGAN. Therefore I changed the code of the original implementation to use celebA instead of mnist. After 15 iterations I am getting the following results:...
1
vote
0answers
50 views

Assymetric filter size on range images in CNN?

I was wondering if there has been any know-how about the filter sizes in CNNs when using panorama images? Would it make sense to stretch the filters the same way as to the ratio of the height and the ...
1
vote
2answers
433 views

To provide dimensionality reduction, 1x1 convolutions are used, before passsing them through a 3x3, or 5x5 convolution in an Inception module.

To my understading what a 1x1 convolution does is gives an embedding of the (i,j)th entry of the feature map along its depth. Besides here some dimensionality reduction is also done. How will the ...
1
vote
0answers
587 views

Different low resolution images using CNN

I am trying to classify different categories of plant using convolutional neural network (CNN). Each category of plant has small set of images. Size of each image is 1000 x 1000. To increase the ...

1 2 3
4
5
7