Stack Exchange Network

Stack Exchange network consists of 174 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

Questions tagged [cnn]

Convolutional Neural Networks (CNN, also called ConvNets) are a tool used for classification tasks and image recognition. The name giving first step is the extraction of features from the input data.

0
votes
0answers
15 views

Transfer Learning - Retraining - 98.9% test and validation accuracy but failing on real data completely

I have around 5000 classes and I retrained a CNN, but it did not classify. So I split the classes and have decided to use multiple nets for batches of 500 classes. I cannot get real images so have to ...
0
votes
0answers
9 views

Convolutional Autoencoder [on hold]

I am new to machine learning and please excuse me if some of my doubts are trivial. a) I am designing a convolutional autoencoder for a problem from computer graphics. The input for this network is ...
1
vote
2answers
12 views

Does fine-tuning of transferred layers perform better than frozen transferred layers?

I recently learned concepts of transfer learning. Is it necessarily true that fine-tuning of transferred layers perform better than frozen transferred layer? why?
2
votes
0answers
18 views

how can I get the original pixels that lead to the decision in CNN ? is that possible?

I work on medical images, I want to locate the most relevant regions of the image based on deep learning spatially CNN, so I feed my data into VGG16 architecture, I get the features maps, now I want ...
1
vote
0answers
3 views

I have to find homography transform parameters which are 8 in number with the help on CNN [on hold]

I have to find homography transform parameters which are 8 in number..I m using a pretrained residual network. The problem I am facing is I am not able to use model.fit() as it takes argument such as ...
2
votes
2answers
21 views

What does the embedding mean in the FaceNet?

I am reading the paper about FaceNet but I can't get what does the embedding mean in this paper? Is it a hidden layer of the deep CNN? P.S. English isn't my native language.
0
votes
0answers
11 views

Combining time-series lstm

I have time series LSTM, which makes fair predictions and might be a first-run model for my needs. An issue that I have is that I have multivariate analysis per user , and so far the LSTM is only ...
1
vote
0answers
9 views

Trainable parameters explode in CNN , how to control them and yet get desired accuracy

I am building a CNN mode using CSV of size (79999, 11) Here is my model below. When I do a model summary it gives me close to two billion trainable parameters , which my laptop or even AWS Ec2 is not ...
2
votes
2answers
39 views

CNN not learning properly

Before marking my question as duplicate, I would like to say that I have tried all the possible solutions mentioned in similar questions, but that doesn't seem to work. I am currently working on ...
0
votes
1answer
23 views

Determining size of FC layer after Conv layer in PyTorch

I am learning PyTorch and CNNs but am confused how the number of inputs to the first FC layer after a Conv2D layer is calculated. My network architecture is shown below, here is my reasoning using ...
0
votes
1answer
7 views

How to obtain and load a good initial data set for object localization?

I'm looking for a good data set for training a CNN based network to do object localization (i.e. a data set with class labels and bounding box data). What is a good initial data set to use? How can ...
0
votes
1answer
22 views

What is the effect of highly correlated data on a Convolutional Neural Network?

A speech audio sample can be converted to MFCC coefficients for further analysis. I wanted to know the effect of correlated data on a CNN. I know the process of computing the MFCC coefficient, which ...
0
votes
0answers
5 views

What would be the advised model that would be needed to be used for human speech recognition?

I needed to build a human speech recognition system which would classify the speech sample into the target classes. I wanted direction as in which deep learning model should be used for the below ...
1
vote
1answer
44 views

image recognition: fully connected network vs CNN

To recognize handwritten digits, I have a fully connected network, containing only 2 layers: input layer (all pixels of the image) and output layer (0 or 1). I use the simplest linear regression for ...
0
votes
0answers
15 views

Ways of estimating data based one a data set

I've a data set which has some physical relationship between them. Let them name as train_x and train_y which both are vectors. ...
0
votes
1answer
19 views

How do I use a model after it's fitted to predict the class of a single string?

After a model is built, how can I use it to predict the class of a single string? model.predict() is returning something like ...
0
votes
1answer
31 views

keras Sequential CNN for image data reshaping data issues

I am new at keras and CNN and am working on building at CNN for sequential analysis of movement in a image. What I am having issues with is the reshaping the data and the labels that go into the ...
2
votes
4answers
48 views

How to prepare the varied size input in CNN prediction

I want to make a CNN model in Keras which can be fed images of different sizes. According to other questions, I could understand how to set a model, like ...
1
vote
1answer
14 views

What exactly does the model generation mean in this diagram?

I've been trying to grasp a research paper on image colorization using neural networks here I am stuck at this diagram. What I need help on, is the Model Generation step after Feature extraction. ...
0
votes
0answers
11 views

Python - Framework for Independent Component Analysis as Neural Network

I'm searching for a method for implementing ICA to my train_x data as roughly like NN. I've train_x and train_y data which both are 1D vectors. Then I want to estimate test_y with test_x On the ...
1
vote
1answer
15 views

What's a difference between neoperceptron and CNN?

Maybe it's a stupid question but I can't get a difference between neoperceptron and CNN. Both ANNs have hidden layers and scanners, as I understood, but many sources subdivides it in two classes. Can ...
0
votes
0answers
15 views

image labeling for training

I am trying to label images for training (will be used in python). I am using mat-lab image labeling app. the output file exported as .mat, how can I export it as .csv as i need to open it in python. ...
0
votes
1answer
43 views

training when Multiple labels per image

I have multiple labels per image. is it better to train taking each each label separately or should i mark all the labels present as 1 in the same image? which method is better? i will be using CNN ...
0
votes
0answers
6 views

Does z-score normalization of the inputs make sense for a binary CNN classifier network?

Assume I have a list of images as a training set with binary labels. I run a z-score normalization to each images $x-\text{mean}(x) \over \text{std}(x)$. The network has activation layers (e.g. ReLU). ...
2
votes
1answer
31 views

Does CNN take care of zoom in images?

Suppose a convolution neural network is trained on small images of an object, say flower, as in following 3 training images: Will this CNN correctly classify if the same object is present in zoomed ...
0
votes
0answers
8 views

How is it that Compute the features by the CNN without targets of train data

Excuse my ignorance, but I am a newbie when it comes to deep learning in general. I am trying to run a training algorithm on train data which is images using VGG16 network part of trained models on ...
0
votes
0answers
12 views

Model for predicting geospatial-temporal (climate) data?

I am attempting to reproduce parts of a dynamic climate model using machine learning in place of physics code. A Jupyter notebook illustrating my current approach is here. I have both inputs and ...
0
votes
1answer
25 views

Complex-Valued input to CNN [closed]

I want to train a CNN. However, my input is images of size 100*100 with complex numbers. I have runned the model, but it failed and the loss didn't decrease. Then I found out that my because my inputs ...
0
votes
0answers
13 views

convolution neural network:representing vector in fully connected layer

please i would like to ask about representing feature vector in the fully connected layer in cnn. i have image and i cropped it into N segments and fed each one into cnn branch and get feature maps ...
0
votes
0answers
22 views

LIME visualization outputs padded regions as important - Mel-spectrogram (audio analysis)

I am encoding audios as Mel-spectrograms and using these Mel-spectrograms as input to my deep learning model (Inception-ResNet V2). The input image is of size 256 X 256, made up of a 128 X 64 ...
0
votes
1answer
61 views

Keras Fit Function (R): Train Regression Model with multiple Labels

I am trying to implement a deep learning model in R using Keras. Let's say I had a dataset of people's faces and a CSV with information about the person's age, gender, and ethnicity. I want to train ...
2
votes
0answers
20 views

it is possible to use features maps of CNN to localised important areas in image?

I'm new in deep learning and CNN, I understand how convolutional and pooling layers work, I understand how and why feature maps are created. How I can localize from the feature maps important area in ...
0
votes
0answers
9 views

Is it worth using residual blocks in a neural network with low number of layers?

I am new to the Deep Learning domain and I was recently reading about the resnet architecture. So I was wondering, can residual blocks improve the performance of even more "shallow" networks, or they ...
0
votes
0answers
45 views

Why CNN and Neural network implementation not working properly

I am working on implementation of Bangla Handwriting Recognition From Scratch. The major steps involved are as follows: Reading the input image. each image shape ( 100,100,3) Number of Train ...
0
votes
1answer
27 views

How filters are made in a CNN?

I am new to Data Science and CNN. My understanding of CNN is that: An image's pixel data is convoluted over with filters which extract features like edges and their position. This creates filter ...
2
votes
1answer
15 views

Relation between amount of training samples and model depth?

When I add more hidden layers to my CNN (e.g. Dense Layers) it seems that the model needs more training samples to produce good results for classes with few training samples. In the single layer case ...
0
votes
0answers
11 views

Quantized fully connected layer implementation issues with VGG16

I recently ported a floating point model of VGG16 to fixed point by quantizing the weights changing the ops to the quantized versions. The network works fine till pool5, and after that the fully ...
2
votes
1answer
40 views

How to analyze CNN model summary and improve it?

I am using a CNN (adapted from a few links on the net) for an image classification task. There are about 8000 images of size 128x128 each. They are of 13 different classes. Following is output of <...
0
votes
1answer
23 views

How to fetch text from pdf to further proceed with question answer based model from the same document?

To illustrate the above title. Suppose you have a pdf document, which is basically scanned from hardcopy, now there are set of fixed questions to answer from the document itself. For an example a ...
1
vote
0answers
14 views

Visualizing mutual information of each convolution layer for image classification problem

I recently came across this paper where the author has proposed a compression based theory on understanding the layers of a DNN. In order to visualize what was going on the authors showed Figure 2 of ...
2
votes
0answers
13 views

CNN kernel location for input image

Given a CNN, say AlexNet: How could one relate kernel locations at the 3rd conv block, i.e 13x13 filter size to the input image. Would that give a meaningful representation in terms of the input ...
1
vote
0answers
26 views

Detect non-digitized university degree's authenticity by Deep Learning [closed]

How can we detect by the look and feel of non-digitized university certificates’ authenticity through machine learning? I am working in this project right now, where I have to detect the authenticity ...
0
votes
2answers
45 views

Relationship between train and test error

I have some specific questions for which I could not extract answers from books. Therefore, I ask for help here and shall be extremely grateful for an intuitive explanation if possible. In general, ...
0
votes
0answers
10 views

Image processing - How to store image list and its labels in one-hot encoded ndarray?

I got a folder with 1976 training images. Each image has a shape (118,128,1) (greyscaled images). I created an array with all the images like this: ...
4
votes
2answers
77 views

Memory error when using more layers in CNN model

On my dell core i7 - 16GB RAM - 4gb 960m GPU laptop, I am working on a project to the classify lung CT images using 3d CNN. I'm using the CPU version of tensorflow. The images are prepared as numpy ...
0
votes
0answers
29 views

Trainig the PNET chain in MTCNN

I am trying to implement the MTCNN as explain in this paper: https://kpzhang93.github.io/MTCNN_face_detection_alignment/index.html I cannot fully understand the training of the cnn chains. Do we also ...
0
votes
0answers
34 views

Increase dataset for training

I have few(~10) csv files of images of high resolution of 4000*300px. I want to train a CNN, using these images. Images comprise of satellite images of craters. Is dividing these large images into ...
2
votes
1answer
39 views

Too little or too much maxpooling?

I am creating a CNN in Keras where model.summary() shows: ...
5
votes
1answer
185 views

An error with respect to filter weights in CNN during the backpropagation

Let's say a convolutional layer takes an input $X$ with dimensions of 5x100x100 and applies 10 filters $F$ 5x5x5, thus produces an output $O$ 10 feature maps 96x96. During the backpropagation the ...
2
votes
1answer
60 views

Python - Predicting data based on multidimensional array with Keras

I've a list of data which is so called 3D array. Each of 10350 rows contains a 2D matrix with size of 150x16 (elements are float) (x_train). Corresponding training data for this huge array a linear ...