Questions tagged [convolutional-neural-networks]
For questions about convolutional neural networks, also known as CNN or ConvNet.
1,159
questions
0
votes
0
answers
10
views
Is size of trained model on disk a good measure of model complexity?
I am writing a research paper on my own custom CNN model for image classification. I am comparing my model architecture with pre-trained architectures, like DenseNet121 and InceptionV3.
I want to ...
- 13
1
vote
1
answer
43
views
Do GANs have constant running time?
After the model is trained, you just need to input random noise and the generator will output an image, does this mean GANs have constant running time ? I'm asking about both naïve GAN and variants of ...
- 31
0
votes
1
answer
21
views
Encoder-Decoder Semantic Segmentation
When doing semantic segmentation, we often make use of FCN's which can be thought of in two parts: an encoder and decoder. As I understand, the encoder compresses the image into a spatially small, but ...
- 101
1
vote
2
answers
146
views
The training process of a conditional GAN
For example, consider a dataset like MNIST. I give the conditional vector to produce only the number $7$ for both the generator and discriminator. In the following scenarios, what will the ...
- 15
0
votes
0
answers
20
views
When to know if I am "on the right track" for a CNN architecture
Context
Very new to CNNs and ML in general. I am building a simple binary image segmentation network for generating black and white image masks (white pixels = desired object; black pixels = all else)....
0
votes
1
answer
35
views
Is it possible to build a convolutional autoencoder with fully connected bottleneck with low dimension?
I want to do a project with a small size image dataset (the size is about 50*50). There's another similar dataset, and I want to prove that the datasets are different. I built a convolutional ...
- 3
0
votes
1
answer
107
views
In the conditional GAN (cGAN) architecture, why does the discriminator need conditional variable?
I'm reading about conditional GAN (cGAN) architecture, what I know is that the generator creates images combining both noise vector and conditional variable, the noise vector brings in random elements ...
- 15
0
votes
0
answers
11
views
How to create meaningful features that allow unsupervised image classification?
I would train features that can later be correlated with categories, once I have some examples for the categories. Let's say one has a set of training images sorted by artist, and wants to create some ...
- 280
0
votes
1
answer
22
views
Generator loss not decreasing while training GAN
I’ve been attempting to create a basic GAN to generate images using this database of flowers (https://www.robots.ox.ac.uk/~vgg/data/flowers/102/).
I’ve spent a few days on this, and largely based my ...
0
votes
0
answers
14
views
How can I feed data files of different sizes to a CNN?
I have 5022 data files. I want to feed them into a CNN model.
However, the lengths of data files vary from, say, 50 to 1015 rows (the number of columns is constant).
In the case of image files, we can ...
- 299
0
votes
0
answers
26
views
Multi-Task VAE, Decoder Activation Functions?
I'm working on a Multi-Task VAE with one Encoder and two Decoders. The input consists of a vector with parameters which describe a design of a fluid system. The goal is to reconstruct the parameters ...
- 1
1
vote
1
answer
41
views
How do I train a CNN on multiple images that have the same shape?
I want to train a convolutional neural network on multiple input images. My image is 240x360 and is in RGB. Therefore my input image has a shape of 3x240x360. Now I want to use multiple images of the ...
- 61
0
votes
0
answers
23
views
In the context of lenet, does C1 refer to the conv layer or the output of the conv layer?
I'm studying lenet.
C1 is the layer
According to a tutorial, C1 is
the first convolutional layer with 6 convolution kernels of size 5× 5.
C1 is the feature map
However, I believe that the part ...
- 231
0
votes
0
answers
23
views
What algorithms could I use if I want to increase the accuracy of matched keypoints in an image pair?
Let's say that I used a keypoint detector like SIFT or SuperPoint to detect keypoints in image 1 and 2. Afterwards, I used a keypoint matcher to match corresponding keypoints in this image pair. The ...
- 101
0
votes
1
answer
78
views
Transfer learning using pretrained tensorflow object detection model [closed]
I am new to AI/ML and wanted to seek guidance as I am totally lost. I will simplify my issue as follows:
Let's say I would like to detect apples and oranges in images.
I would like to leverage a pre-...
- 3
0
votes
0
answers
18
views
Discussion about Improving Visual Search Model Accuracy
My Visual Search Model is only achieving an accuracy of about 42% If anyone can give me advice to drastically improve this number I would greatly appreciate it. Below is my current flow of image ...
0
votes
0
answers
12
views
Possible Reasons for the Discrepancy in Trainable Parameters of the Extended DeepConvLSTM Model
I have implemented DeepConvLSTM baseline Model input are 60×d frames each representing 60 samples with d features. Frames are fed into four consecutive convolution layers with standard rectified ...
- 41
1
vote
0
answers
22
views
Siamese network, cosine similarity unexpected result?
I was reading more about siamese network and it's use for similarity problems and I've stumbled upon this https://keras.io/examples/vision/siamese_network/
I was surprised to see both similarities in ...
- 11
0
votes
0
answers
20
views
How to compensate the receptive field offset in a Fully Convolutionl Network?
I'm studying the Fully Convolutional Networks right now and when it's clear that the receptive field is not dependent on the input size (the whole network in a way is independent from the input size), ...
- 1
0
votes
0
answers
32
views
How does a zero-order hold kernel in a Convolutional Neural Network look like?
Several papers co-authored by Hitoshi Kiya propose to use a fixed convolutional layer with a zero-order hold kernel to avoid checkerboard artifacts in CNNs. [1, 2, 3]
While there is plenty of ...
- 101
0
votes
0
answers
13
views
Feeding variable length of 2D image slices of the MRI into the deep neural network
I am trying to build a classifier that would predict the correct outcome (disease vs healthy) using a set of 2D slices derived from the 3D MRI scan. For each patient, based on the 3D scan, I am able ...
- 101
0
votes
0
answers
13
views
What are some good pairs of transfer learning source and target datasets for image classification?
As the title says, I'm curious about some well used transfer learning tasks.
ImageNet to other datasets is common, but what are something good pairs I can try and mess around with ?
Like CIFAR10 to ...
- 11
0
votes
0
answers
41
views
VAE for Motion Sequence Generation - Convergence Issue with Scheduled Sampling
I have implemented a Variational Autoencoder (VAE) in PyTorch for motion sequence generation using human pose data (joint angles and angular velocities in radians) from the CMU dataset. The VAE ...
- 1
-1
votes
1
answer
42
views
What is the best lightweight alternative to VGG16 for image fingerprinting?
I am using a VGG16 model with the classification layer stripped off to generate vectors for an intermediate stage of an image fingerprinting algorithm. It works well, but VGG16 is a little hefty, and ...
- 136
0
votes
0
answers
35
views
Why are convolutions and pooling described as layers in a network?
Whenever I look at resources on convolutions and max-pooling in CNNs they always seem to describe these algorithms as being part of the network - a preliminary set of layers before the main ...
- 13
0
votes
0
answers
29
views
Is my 1D signal using CNN & RNN regression reasonable?
I want to know if my impact-echo signals are proper with CNN or RNN regression model.
I got some simulated signal, as following shows.
In previous research, people mostly consider frequency or even ...
- 1
1
vote
0
answers
39
views
When training a convolutional neural net, what if the channels in my image are mutually exclusive?
I'm attempting to train a convolutional neural net to perform binary classification of volumes of shape $(H, W, C)$ (i.e., height, width, channels). For the sake of this example, let's say that the ...
- 11
0
votes
1
answer
25
views
Can a concept/feature be represented using more than one layer of a Neural Network?
I was reading Goodfellow. At the start of the text it was mentioned that there are two ways to represent depth of a deep neural network. One is using the depth of the computation graph and the other ...
- 101
2
votes
2
answers
57
views
Is it possible to reconstruct convolutional layers' input using transposed convolution?
I've been trying to visualize internal activations in CNN and came across this paper: "Visualizing and Understanding Convolutional Networks" by Zeiler & Fergus.
In the paper they ...
- 143
0
votes
0
answers
22
views
Loss function increases dramatically after transfer learning (or parameters initialization)
I am modeling a CNN network with some customized layers that performs 2D FFT and IFFT, I have a dataset with features representing time domain OFDM symbols, and labels representing frequency domain ...
0
votes
1
answer
40
views
Image segmentation with varying resolution
I am looking to create a model that is able to perform binary segmentation of images with varying resolutions. For model should be able to classify tree or not tree regardless of the resolution of the ...
- 101
0
votes
1
answer
41
views
How to add engineered features to an image segmentation model
I have built a U-net model for image segmentation of 3-channel remote sensing images. I have a total of four classes; two of these classes look very similar and are hard to distinguish in the images ...
- 1
2
votes
2
answers
52
views
How translation invariance is achieved in CNNs?
I am trying to understand how translation invariance is achieved in CNNs. For example, consider the following simple binary classification problem: predicting whether the letter that appears on an ...
- 150
0
votes
1
answer
42
views
Temporally Non-Aware RNN
I am trying to classify whether or not a specific object is in panoramic photos. The issue is, a panoramic photo can be any width, so the input to my neural network can't be fixed in that dimension.
I'...
user55176
0
votes
0
answers
11
views
Neural Network (NN) to predict the duration (in seconds) of a fault, how to recognize the elapsed time in prediction of an active fault?
I'm working on developing a neural network (NN) to predict the duration (in seconds) of a fault. However, I've encountered a couple of challenges:
Model Performance: My neural network seems to be ...
- 1
0
votes
0
answers
14
views
NN architecture for a sparse pixe-wise image2image mapping
Hi,
I have pairs of images like below (inputs and labels), they are both single channel images with real-valued pixels. I want to train a CNN that maps from one to the other. I have tried UNet and ...
0
votes
1
answer
76
views
Why don't people use their own random noise to counter adversarial attacks on computer vision systems?
Why couldn't you take the image an AI is given and apply several different random noise filters to the image and take the democratically most common response and use that for the output of the AI. As ...
- 103
2
votes
0
answers
31
views
What's the result of multiple neurons using ReLU activation function?
This question comes from a doubt that I recently had on an amazing book called "Neural Networks from Scratch With Python by Harrison Kinsley & Daniel Kukieła"
Let's suppose that I have ...
- 21
0
votes
1
answer
33
views
YOLOv1 doesn't work for custom dataset
I am currently trying to train my own YOLOv1 network, based on this repository:
https://github.com/ivanwhaf/yolov1-pytorch
The images I want to train look like this:
Three classes I want to detect:
...
- 111
1
vote
1
answer
48
views
Convolution layer, with biases too?
We already know that the kernel slides around the image, multiplying the pixels with the parameters, so, what if additionally, we also have a kernel slide around the image and add values(different ...
- 11
0
votes
1
answer
52
views
CNN how to measure the amount of FPS that can be processed?
This is my first question in the AI stack exchange. I want to ask about how to measure how many FPS can a CNN model process during real time detection. I am working on a real time detection system ...
- 3
1
vote
1
answer
62
views
Dealing with noise in models with softmax output
I have a device with an accelerometer and gyroscope (6-axis). The device sends live raw telemetry data to the model 40 samples for each input, 6 values per sample (accelerometer xyz, gyroscope xyz). ...
- 113
1
vote
1
answer
35
views
Number of units in Final softmax layer in VGGNet16
I am trying to implement and train neural network model VGGNet from scratch, on my own data. I am reproducing all the layers of the model. I am having a confusion about the last, fully connected ...
- 13
1
vote
1
answer
34
views
CNN-Regression insensitive to input data
I'm currently training a CNN + multiple target regression model that does the following
input: $ \dim x = (L, 2), \text{where} \ x_i \in (-0.1, 0.1) $
output: $\dim y = (M), \text{where} \ y_i \geq ...
- 11
0
votes
0
answers
53
views
How do i approach creating a masked auto-encoder for feature extraction
I trained Masked Autoencoder-based models in order to use the encoder as a backbone for another task. Pretraining has been done in a Self-Supervised manner on image data. Now that it comes to my ...
- 1
0
votes
0
answers
21
views
Binary classification using softmax and categorical crossentropy: monitoring validation
I run a binary classification using different CNN versions in Tensorflow.
When I label samples from each class using 0 and 1, I select a sigmoid output in the last layer of the CNN, like
...
- 101
0
votes
0
answers
51
views
How to improve CRNN model(CTC loss) accuracy for OCR task?
I take this as baseline model. The main difference on RNN part,
...
- 101
0
votes
2
answers
39
views
How is the number of channels in a convolutional layer shrinked or expanded?
I know in order to shrink or expand the number of channels a 1x1 convolution is performed.
I need to clarify the following: is the 1x1 convolution(s) just a matrix multiplication between the image ...
- 101
0
votes
0
answers
15
views
Quantization Parameters when converting Quantized Transposed Convolution to Conv2D
A simple way to compute TransposedConv2d is to convert it to a regular Conv2d by padding the input value with zeros, as is described in A guide to convolution arithmetic for deep
learning. Does this ...
- 3
0
votes
1
answer
51
views
How does Conv2Plus1D reduces the number of paramateres?
Based on this tutorial, and the
An advantage of this approach is that factorizing the convolutions into spatial and temporal dimensions saves parameters.
statement, the Conv2Plus1D must have fewer ...
- 69