All Questions
Tagged with convolutional-neural-networks computer-vision
139 questions
2
votes
0
answers
18
views
How can I visualize pixel level information perceived by a model (VGG-16)?
As a fast.ai starter project I regressed a little model on movie stills from various eras to see if I could predict a year of release given an unseen image from a film. It works reasonably well! (feel ...
- 121
0
votes
1
answer
69
views
Fine-tune vs training from scratch
I'm training a 2 class Yolov8 Small detection model, and iterated through the model re-training over the previous best model a few times. Now, I added more data and the size of dataset is 2x (200,000 ...
- 101
0
votes
0
answers
21
views
What is "Explicit Propagation" Image Inpainting - LatentPaint Paper - Generative AI
I am trying to implement "Explicit Propagation" method for Image Inpainting introduced in LatentPaint Research paper.
This paper proposes the introduction of a module between VAE and ...
0
votes
0
answers
12
views
Understanding matching of a CNN Layer's Output With the Receptive Field of Input Layer
I was trying to implement the following paper: https://arxiv.org/abs/1610.01563 and I came across something that seemed ambiguous to me. On page 4, second paragraph, it says
After processing the ...
0
votes
0
answers
62
views
Trained model on cifar10 performs poorly on real images
So I'm trying to train a model using the CIFAR10 dataset.
The problem is that while the performance of the model on validation and test sets are good (about 95-96%), the model fails to predict images ...
4
votes
2
answers
1k
views
Why different noise in GAN generate different images?
I understand that noise $z$ serves as the input to the generator. Noise $z$ is essentially a vector of random numbers, typically from Gaussian distribution with chosen size of like $100$. However, I ...
user77925
0
votes
0
answers
22
views
adversarial training on convnext shows a very strange curve
i am currently working on a research project where I have to train some models for adversarial robustness. I have implemented the algorithm used by a research paper called adversarial training for ...
- 1
1
vote
0
answers
55
views
Classifying Images that Look Like Noise
I'm about to build a system that is supposed to evaluate images (900 x 150) like the following and classify it in to one of five categories:
image that looks like noise
In case you're wondering, they'...
- 11
0
votes
0
answers
193
views
How does the Yolo loss match bounding boxes to the ground truth?
I've been going over the YOLO paper again and I was wondering something about the loss.
Yolo divides an image into grids and then for each grid can have multiple bounding boxes to detect multiple ...
- 101
1
vote
1
answer
115
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 ...
user78202
1
vote
2
answers
259
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 ...
user77925
0
votes
1
answer
131
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
386
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 ...
user77925
1
vote
0
answers
89
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
-1
votes
1
answer
356
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 ...
- 156
0
votes
1
answer
76
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
0
votes
1
answer
130
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
0
votes
2
answers
116
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
1
vote
0
answers
52
views
What is the number of channels of input audio mel spectrogram?
What is the number of channels of input audio mel spectrogram? For example, in CV we always have 3 input channels on RGB picture. But what about audio?
2
votes
0
answers
65
views
Is it possible for original Vision Transformer (ViT) to do fine-grained semanantic segmentation? if so, how?
As far as I know, in the original ViT, the image is first divided to a fixed size of patch (16x16, for example) then they are flattened and treated as tokens and fed into Transformer.
Without using ...
- 21
1
vote
0
answers
11
views
Filter in the Single Shot Detector
Let's say I want to implement a single shot detector. When I get a feature mal as an input, I will use a 3x3 filter for prediction for each cell.
Let's say we have 5 classes with 6 ancors, I would ...
1
vote
1
answer
249
views
Are CNNs exactly translation invariant with global max/average pooling layers?
CNNs are naturally translation equivariant, meaning that if we translate the input, then the feature maps are translated the equally.
With the use of max/avg pooling layers, this translation ...
0
votes
0
answers
35
views
CNN without actuators
After training CNNs without actuators, I have an idea to compare their weights with each other using image mirroring. I am looking for ideas about reality perception of CNNs in this way.
What might ...
- 1
1
vote
1
answer
325
views
Are there any advantages of encoding an image as a graph to use in Graph Convolutional Networks?
I have seen this encoding of an image as a graph:
The set of the nodes $V$ is the set of pixels. If the image is of size $10\times10$, then we have $10\cdot10=100$ pixels.
Each node has a length 3 ...
3
votes
0
answers
62
views
Why do adversarial attacks work on CNNs if they classify images as humans do?
A common illustration on how CNN works is as follows: https://www.researchgate.net/figure/Learned-features-from-a-Convolutional-Neural-Network_fig1_319253577. It seems to suggest that CNN in ...
- 195
0
votes
2
answers
155
views
How can I use CNN to make a cumulative count of the number of occurrences of each of the different objects in all the images in the test set?
Let's say there are three images in the test set, the first with three triangles, the second with two triangles and two circles, the third with four circles and two squares, and the final tally is a ...
- 1
0
votes
1
answer
43
views
How are OCR training datasets constructed?
For the sake of concreteness: let's suppose that the word "OCR" refers to any OCR system build on an R-CNN architecture. Similarly, in aims of simplicity, let's declare that we are ...
- 133
0
votes
1
answer
74
views
Is there a best practice for creating multiple convolutional layers from small image inputs?
With all the work being done on larger and larger images, I'd like to ask if a best practice(s) has arisen for allowing multiple convolutional layers on small image inputs?
For instance, in my case I ...
- 99
0
votes
0
answers
553
views
How to increase accuracy for CNN?
I have built one CNN model and applied it to chest-xray Covid 19 pneumonia dataset. I am getting the classification report as follows:
I am surprised to see that it is giving an excellent result on ...
- 121
0
votes
1
answer
47
views
How to use strong labels in image classification?
I have a dataset where I have the labels cancer & non-cancer, and I also have localized pixel-level annotation masks of important regions/features in the images.
In a binary classification task, ...
- 11
1
vote
0
answers
39
views
How does a CNN work in detecting absence of features?
I'm trying to understand how a CNN operates internally. Let's say I'm doing binary classification with 1 output neuron and a sigmoid to classify dog vs no dog. No dog meaning the image does not ...
- 11
1
vote
0
answers
47
views
Help with model architecture for a racing game
I’m working on a model for a racing game using pytorch. The model gets frame from the game as input and produces a controller state as output. The dataset consists of frames from the game and ...
- 11
1
vote
0
answers
37
views
Why is my convolutional neural network failing to classify user inputted images after having high accuracy in testing? [closed]
My CNN was trained on the Kaggle A-Z Dataset and consists of:
...
0
votes
1
answer
749
views
Training strategy on continuous video stream with CNN-LSTM
I have videos that are each about 30-40 mins long. With the first 5-10 mins (at 60fps, can be down-sampled to 5fps) are one type of activity that would be categorized by label-1 and the rest of the ...
- 101
1
vote
1
answer
112
views
Does Using the Same Background for Binary Classification Improve Model Accuracy?
I am training a CNN that detects if a there is a pot of boiling water vs if there is a pot of boiling water with pasta inside. My hypothesis is that having the same background for both a positive and ...
- 123
0
votes
2
answers
130
views
Are there any works that deal with 2D pose estimation in videos?
Since pose estimation is often a task where spatial-temporal context should be helpful in finding subsequent key points, I thought there should be many papers on it. However, I could not find any work ...
- 635
0
votes
0
answers
164
views
Predict placement of an object in 3D space
I am trying to find a way to train a model to predict the correct placement of entities like a tree, dog and cat in a natural 3D environment. Any help regarding how I could use textual data to learn ...
2
votes
1
answer
380
views
CNN Architectures for local features vs global context
Kaparthy in his blog post said
[this] hints at the kinds of architectures we’ll eventually explore. As an example - are very local features enough or do we need global context?
I'd like to gain ...
- 366
1
vote
0
answers
127
views
Hand Landmark Detector Not Converging
I'm currently trying to train a custom model with TensorFlow to detect 17 landmarks/keypoints on each of 2 hands shown in an image (fingertips, first knuckles, bottom knuckles, wrist, and palm), for ...
- 111
2
votes
2
answers
92
views
How general is generalization?
I am sorry but I have to explain my question using an example, I do not know how to ask it in proper scientific terms.
Let's assume, I have trained a deep learning model on classifying hand gestures, ...
- 1,280
0
votes
1
answer
58
views
How to divide a segmented image into classes instances?
Is there a method/algorithm to generate instances of objects from image that was segmented by the use of any image segmentation models?
For example, I have an image with one class and it was segmented ...
- 1,280
0
votes
1
answer
400
views
Is it possible to use deep learning to generate a 2D image from a few numerical values?
Is it possible to train a DL model that will generate a full resolution 2D image based on few numbers describing this image and what type of model or architecture would that be?
What I want to ...
- 1,280
0
votes
0
answers
50
views
Does randomly adding hand-engineered features increase the CNN's sample efficiency/performance?
It is a known fact that preprocessing images using CV techniques will improve CNN performance (see this answer).
But what happens when you feed in the entire image and the filtered image randomly to ...
- 670
0
votes
0
answers
68
views
Loss function to Push response value towards extremes
I have a feature map whose values are in the range of [0,1]. I want to push these values either towards extreme 0 or 1 using some loss function. Since I don't have any target value so it had to be in ...
0
votes
0
answers
19
views
How to change a single object detection network to a multiple object detection network?
I have trained a CNN network to detect a circle and approximate its centre and radius in an image. What I want to do now is detect the centre and radius of all the circles if there are multiple ...
- 133
2
votes
1
answer
121
views
Is intersection of labels acceptable in computer vision?
I have a dataset, where objects are very close to each other. So, the question is: what is the best approach to label them?
There are two possible options:
mark objects so that they will not ...
- 121
1
vote
1
answer
2k
views
Should one use an "other" category in image classification?
In image classification, there are sometimes images that do not fit in any category.
For example, if I build a CNN in Keras to classify Dogs and Cats, does it help (in terms of training time and ...
- 113
2
votes
0
answers
59
views
What are the metrics to be used for unsupervised monocular depth estimation in computer vision?
I am currently replicating the results of this paper. In this paper they have not mentioned how they are evaluating the results as no ground truth is available for comparison. Same goes for other ...
- 61
2
votes
0
answers
43
views
CNN leaf segmentation throught classification of edges how to improve
I am trying to design a CNN that can do pixel wise segmentation of edges leaves in dense foliage agriculture images. Such as these:
On the basis of this article https://arxiv.org/pdf/1904.03124.pdf, ...
- 21
3
votes
1
answer
521
views
How do you calculate KL divergence on a three-dimensional space for a Variational Autoencoder?
I'm trying to implement a variational auto-encoder (as seen in Section 3.1 here: https://arxiv.org/pdf/2004.06271.pdf).
It differs from a traditional VAE because it encodes its input images to three-...
