Questions tagged [convnet]

For questions regarding "Convolutional Neural Networks" (CNN)

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

Input 0 is incompatible with layer conv1d_40: expected ndim=3, found ndim=2

i am working on computer vision using deep learning. my training data contains (x,128) shape. i am passing the same to conv1d layer but facing issues below is my model ...
3
votes
1answer
29 views

Triplet loss function for face recognition?

In the Andrew-NG coursera course on Convnets he talked about triplet loss function for one shot face recognition. The formula given in the video is, $$\to \small \small \small ||f(A)-f(P)||^2 \;+\;\...
0
votes
1answer
16 views

How many parameters in a Conv2d Layer?

I was following andrew-ng coursera course on deep learning and there's a question that has been asked there which I couldn't figure out the answer for? Suppose your input is a 300 by 300 color (RGB) ...
0
votes
0answers
21 views

How to design my own keras layer?

I am implementing the paper Perceptual GAN for small object detection. The design is described by the picture given below. I need to design my own keras layer. I have described my code below: The ...
0
votes
0answers
11 views

why averaging the predictions of 4 models , returns a bad result?

I am working on a multi class classification with eight (8) classes. I tried something at the beginning I thought it was a great idea, but it return utter trash. the experiment: I trained each two ...
1
vote
1answer
24 views

Making sense of indices in 2D convolution operations in convolutional neural networks

Referring to the answer here: https://www.quora.com/Why-are-convolutional-nets-called-so-when-they-are-actually-doing-correlations, the equation for a discrete 2D convolution is specified as: $$C(x,y)...
0
votes
0answers
7 views

C3D feature extraction of a video and using PCA to reduce dimension to 500

can someone provide me the code for C3D feature extraction of a video free from caffe installation and using PCA to reduce dimension to 500 as i am facing difficulty because caffe is not getting ...
0
votes
2answers
22 views

Keras Conv1D model Input_shape value error

I am not sure why I am receiving this value error. Additionally, I haven't found a tutorial that explicitly talks about the appropriateness of size of filters and kernel. I would appreciate some input ...
1
vote
1answer
25 views

conv net data retrieval on unseen class

I have build a conv net for image classification which work "well" Now I extract features from last fully connected layer and use it for image retrieval (find image most similar to my target image) ...
0
votes
1answer
10 views

Reason for Training and test loss sudden increment after some epochs keras

We know that if training and test loss are different from each other, our model is over-fitting. However, if both get high after some epochs, how can we justify it? One way to solve it is to reduce ...
1
vote
1answer
41 views

Strange binary classification result with a model that indicate it has been well-trained

The problem : I am trying to build a model for binary classification for melanoma 'MEL' and nevus 'NV' the dataset is from ISIC archive ISIC 2019 but for 8 different type of skin lesion, I am using ...
5
votes
3answers
60 views

How can I know if my conv1D model is overfitted or underfitted from loss curve?

I am working on classification of time series multivariate data. By doing PCA, I converted multivariate to uni-variate and fed it into a conv1d in keras. However, I am getting a very high accuracy ...
0
votes
0answers
11 views

What is the initial size of anchor in Faster R-CNN?

So we generate anchors for input images which will be later used for classification and then regression for bounding box. If we have image size of 224*224*3 and our feature map is of size 7*7*512. ...
0
votes
0answers
8 views

Object Detection using Faster R-CNN conundrum

So we have our image right? We use some pre-trained model like VGG or Inception which will predict the feature_map. Suppose to a shape of (7,7,512) from the original of (224,224,3). We use this ...
1
vote
0answers
25 views

TensorFlow: how to restore pre-trained meta model and pass it's weights and biases to the optimizer?

I trained a model on a specific dataset and saved it as a meta, I want to restore the model and use its weights and biases on another dataset the code isn't mine but I'm trying to restore the ...
0
votes
0answers
6 views

Coreml model inference speed

Comparing the inference speed of my style transfer model on iPhone X vs iPhone 7 vs iPhone XS. Strangely, but the number of frames per second that the model is capable of stylising is pretty much the ...
0
votes
0answers
21 views

Help understanding if suffering from Validation Bias

The goal is to forecast the volume a product will sell in future months. There are about 107 products that are being bought by different customers for different uses. It is univariate problem since ...
0
votes
0answers
9 views

NCHW vs NHWC in Machine Learning

As I've been introducing myself to the various deep learning frameworks, I've noticed a difference in the default placement of channels for images. Is there a substantial difference between NCHW vs ...
1
vote
1answer
45 views

Overfitting CNN models

I tried to develop a number of CNN architectures to train on a 1000-point subset of the "cat-dog" Kaggle training set (meaning, by the way, that all 1000 data points were labeled). I used a 700-150-...
0
votes
0answers
25 views

Image anomaly detection

I am using a conv net to classify 20 different patterns on an image. My train/test set are images where I know the class. With this, I a good enough model for my preference. It's an unbalance dataset, ...
0
votes
2answers
24 views

Unstable results when I train a CNN

I'm currently training a CNN to do a binary classification. I'm getting fairly good results, but unfortunately the training is very unstable. Just by changing the seed the relative error changes by 20-...
0
votes
0answers
9 views

Open-Set Classification

How can we extend the training process of a classification model to be aware of an open-set task? More plainly speaking, is there a way to train a model on a training set that contains samples from x ...
0
votes
0answers
6 views

In a CNN do filters that account for depth do so for just the initial input, or for a layers with any depth?

For example, if you have an input of an image of size 100x100x3 (where 3 is the RGB of the image), then put it through a conv layer that results in an output of <...
0
votes
0answers
15 views

Calculating the Number of Parameters of a 2D CNN Layer

How can I calculate the number of parameters for a 2D CNN layer? I usually use the equation: output width= ((W-F+2*P )/S)+1 = (x) The same answer will be valid for the output height considering that ...
2
votes
1answer
24 views

How to perform polynomial landmark detection with deep learning

I am trying to build a system to segment vehicles using a deep convolutional neural network. I am familiar with predicting a set amount of points (i.e. ending a neural architecture with a Dense layer ...
0
votes
0answers
36 views

Label embedding in Auxiliary Classifier GANs

In Auxiliary Classifier GAN the generator takes two inputs, 1. one hot encoding of the labels, and 2. noise vector. But in the implementation of the GAN (e.g.:) some embedding is used, I think it is ...
1
vote
0answers
60 views

What is the best architecture for Auto-Encoder for image reconstruction?

I am trying to use Convultional Auto-Encoder for its latent space (embedding layer), specifically, I want to use the embedding for K-nearest neighbor search in the latent space (similar idea to ...
1
vote
1answer
50 views

Accuracy improving but, val_acc oscillating in ConvNet. What does it mean?

In my ConvNet model, i'm trying to classify some images. It is malware images and it doesn't contain complex features (i think), as expected model learn to classify images easily. You can see my ...
2
votes
1answer
338 views

What is fractionally-strided convolution layer?

In paper Generating High-Quality Crowd Density Maps using Contextual Pyramid CNNs, in Section 3.4, it said Since, the aim of this work is to estimate high-resolution and high-quality density maps,...
1
vote
0answers
130 views

Character segmentation using deep learning

I'm developing a character segmentation algorithm for license plate OCR. My algorithm includes two steps: segmentation and recognition. There is almost no problem for recognition thanks to CNN. My ...
1
vote
1answer
42 views

Why can't I use data augmentation with a pretrained convnet?

Reading Deep Learning with Python by François Chollet. In section 5.3.1, we've instantiated a pretrained convnet, VGG16, and are given two options to proceed: A) Running the convolutional base over ...
0
votes
0answers
38 views

Dimension of feature vectors for classification task in the DCGAN paper

I am trying to implement one of the section in the DCGAN paper (https://arxiv.org/pdf/1511.06434.pdf) i.e. Using the Discriminator network trained on ImageNet-1k as a feature extractor to classify ...
1
vote
0answers
17 views

Paper about AE-CNN is unclear. Deriving layers of dense blocks?

I am implementing the algorithm called Automatically Evolving CNN (AE-CNN). Some things aren't specified which makes it a bit hard to understand what the paper actually means to say. In the chapter 3....
0
votes
0answers
12 views

How to custom build a convolution generator network?

I learned GAN's by using mnist dataset(28x28) and codes available in the web. Now I am trying to build a GAN for dataset with images containing custom channel, rows, columns. eg:(3,300,200). I have ...
0
votes
1answer
33 views

Setting input shape for an NLP task in R(Rstudio) using keras 1D convolution layer, when it expects 3 dimensional input (a tensor)

I am using R programming language and using Keras API to build a functional 1D CNN. I have a matrix of my dataset of the following shape rows*features (6000*1024). The input layer is set using the ...
0
votes
1answer
54 views

Weighted samples in Tensorflow for convolutional neural networks

For my binary classification problem (A vs B), each image in either class has its individual weight. This means, for example, if I have 10000 images for A, not all of the images are equally important....
2
votes
0answers
397 views

Multi label classification and sigmoid function

I'm new to neural networks so this may be silly question. I have build standard CNN network for image classification. I want multi-label classification network so I use binary_crossentropy as loss ...
0
votes
1answer
64 views

Application of Deep Reinforcement Learning

I'm new to deep learning, and especially to reinforcement learning. I would like to know if it's possible to predict which combination of hashtags (from a subset of chosen hashtags) would produce the ...
0
votes
1answer
114 views

Conv bias or not with Instance Normalization?

It is well known that Conv layers that are followed by BatchNorm ones should not have bias due to BatchNorm having a bias term. Using InstanceNorm however, the statistics are instance-specific rather ...
0
votes
0answers
87 views

Siamese network using VGG16 to verify the similarity images

I want to create a deep learning model which verify the similarity of the images. So, I will use Siamese network. My dataset is images dataset not CSV file. How can I create a Siamese network using ...
4
votes
1answer
79 views

CNN computing time on good CPU vs cheap GPU

I am a researcher working on my first deep learning project, which consists of using a CNN (pre-trained VGG16+2 densely connected layers) to classify drone imagery of vegetation. In trying to hack ...
0
votes
0answers
40 views

Equivariant Representation in CNN

When I was looking through the internet about the explanation of equivariant representation, I found out that a function is said to be equivariant if and only if g[f(x)] = f(h[x]), where g and h are ...
2
votes
0answers
163 views

1x1 convolutions, equivalence with fully connected layer

I'm confused by the concept of equating a 1x1 convolution with a fully connected layer. Take the following simple example of a 1x1 convolution of 2 input channels each of size 2x2, and a single output ...
1
vote
0answers
17 views

Supervisory information through side output in convolutional neural network

I am trying to implement this paper https://ieeexplore.ieee.org/document/7828014 Here they have mentioned text local (edge) and global regions as supervisory information. Side output is generated ...
0
votes
0answers
24 views

Can we do convolutions on binary mask inputs?

I am training a vehicle trajectory prediction algorithm using Deep MaxEnt Inverse Reinforcement Learning (https://arxiv.org/abs/1507.04888). My intention is to have as input to this algorithm a top-...
1
vote
1answer
148 views

Can I use the Softmax function with a binary classification in deep learning?

I want to create a deep learning model (CNN) for binary classification, can I used the softmax function instead of the sigmoid function in binary classification? Adding the classification layer to ...
0
votes
0answers
116 views

How to project a bounding box on feature map?

I'm trying to implement a custom ROI pooling layer in Keras. According to the Fast-RCNN publication, ROI pooling is done this way: RoI max pooling works by dividing the $h \times w$ RoI window into ...
0
votes
0answers
101 views

LeNet-5 - combining feature maps in C3 layer

Famous LeNet-5 architecture looks like this: The output of layer S2 has dimension: 10x10x6 - so basically an image with 6 convultions applied to it to derive features. If each dimension was again ...
0
votes
0answers
33 views

How to train two neural networks together

This could be considered as an extension of my previous question "How to make a region of interest proposal from convolutional feature maps?". Network 1: I have a multi-input neural network, it ...
2
votes
2answers
442 views

Can pooling ever increase accuracy in convolutional neural networks?

In ConvNets, pooling is used to downsize the input volume, leading to fewer parameters, leading to computational efficiency and possibly helping with overfitting. But can pooling ever increase the ...