Questions tagged [data-augmentation]

The tag has no usage guidance.

Filter by
Sorted by
Tagged with
1
vote
1answer
26 views

Is there any possibility to apply deep dreaming in data augmentation?

I looked into the deep dreaming concepts and feel like this has the potential for data generation. But i'm not sure how possible this concept is. Any thoughts regarding this?
0
votes
0answers
15 views

How to change voice features in python without affecting speech/language features?

I am trying to build a CNN model which should be able to identify the language being spoken in an audio file. I have extracted the MFCC matrix (for 13 coefficients) for each audio file and trained it....
0
votes
0answers
6 views

How to use ImageDataGenerator for uint16 images

Keras does not support png uint16 images and convert them into uint8. I found some changes made here, but not adopted yet in the installed version in Google Colab. I wonder if there is any way to ...
1
vote
2answers
28 views

Problem of continuous training - Supervised learning

I am sure this is a most common problem, but would like to know by experts on how to tackle it. Note that, I mostly deal with textual data (NLP problems). When a supervised learning model is created, ...
0
votes
0answers
32 views

How to measure augmented data quality

I work on NLP binary classification task (but actually the question can be applied to any ML task using augmentation) and use Augmentation technique for creating additional data. I already have ...
4
votes
2answers
201 views

Why are results without Transfer Learning better than with Transfer Learning?

I developed a neural network for license plate recognition and used the EfficientNet architecture (https://keras.io/api/applications/efficientnet/#efficientnetb0-function) with and without pretrained ...
0
votes
0answers
48 views

Data augmentation for recommendation systems

I have a user-item matrix that I use to train a denoising autoencoder to predict the top-k items to recommend to the different users. The idea is to corrupt the matrix by erasing a percentage ...
1
vote
1answer
34 views

Training with less data

Most problem with machine learning projects I have faced is the lack of data. The samples available are enough to disqualify rule based approach but not enough for a neural network to train. For ...
0
votes
1answer
62 views

How to train a keras model on both original and augmented data from ImageDataGenerator?

I have a dataset that contains about 87000 images in a directory, with each class in a separate subfolder. I've tried the class ImageDataGenerator() and the ...
1
vote
2answers
36 views

Should augmentation also be performed on the validation set when the dataset is imbalanced?

I am training a CNN on images (2 classes) and I have an imbalanced dataset (1:7 ratio). I am trying to tackle this by performing offline image augmentation. Should I perform augmentation also on the ...
0
votes
1answer
27 views

What's the difference between keras api augmentation and data augmentation definition?

The augmentation definition is increasing the number of images by using rotation, crop and flip to avoid overfitting. The keras API apply augmentation but no increasing the number of image. What keras ...
0
votes
0answers
19 views

Why does the model make good predictions for augmented images and not for the original ones?

I am training a CNN using maps images and I have performed offline augmentation operations (FLIP_LEFT_RIGHT, FLIP_TOP_BOTTOM, ROTATE_90, ROTATE_180, ROTATE_270, TRANSPOSE, TRANSVERSE) on the whole ...
2
votes
1answer
33 views

How to do data augmentation for Machine Learning on statistical data?

I am training machine learning classification model. I have data in csv format with lets say 5 features(or columns) and 100 such observations(or rows). I want to add more similar data to improve my ...
1
vote
0answers
38 views

What is the difference between domain randomization and data augmentation?

Domain randomization (https://arxiv.org/abs/1703.06907) is used to create a synthetic dataset with enough variance that it will encompass unseen real data, as just one variation. I am trying to ...
0
votes
1answer
28 views

What is the principle of Unsupervised Data Augmentation (UDA)? Why does UDA work?

UDA(https://github.com/google-research/uda) could achieve good accuracy by only 20 training data on text classification. But I find it is hard to reproduce the result on my own dataset. So I want to ...
0
votes
0answers
28 views

Problem with data augmentation only on train set!

So I want to augment my data but unfortunately, I don't have an individual folder for train and validation set, and of course, I want to only augment the train set but if I use this: ...
0
votes
1answer
221 views

Data augmentation for multiple output heads in Keras

I have a transfer learning based two output classification problem. So, accordingly, I have formatted my data to have X_train as a ...
0
votes
0answers
22 views

Time Series Model (n-beats paper) how is sampling / training done?

does any one know what this means? It is taken from the paper https://openreview.net/pdf?id=r1ecqn4YwB (n-beats time series model). To update network parameters for one horizon, we sample train ...
1
vote
1answer
58 views

How can I improve my model on a very very small dataset?

I am starting as a PhD student and we want to find appropriate materials (with certain qualities) from basic chemical properties like charge, etc. There are a lot of models and datasets in similar ...
1
vote
1answer
20 views

After performing data Augmentation on tf.data.Dataset, should i MERGE it with original tf.data.Dataset?

Kind of a silly question, but I read that data Augmentation can be used in order to solve problem of small datasets. In my case, I've got a dataset with 5 different classes and around 2k examples per ...
1
vote
0answers
54 views

Flow_from_directory() won't load any image

i'm trying to load images for training a CNN using keras flow_from_directorY(), but seems that is having hard times finding the images. i'm running the code on google colab ...
1
vote
0answers
459 views

ValueError: could not convert string to float: 'Nor967.jpg'

Whenever I try to use the data augmentation ImageDataGenerator I'm getting an error like could not convert string to float. Here is my code. ...
0
votes
1answer
42 views

Why do we use different image processing at train and test time for an image classification tasl

I would like to ask why we use different image processing at train and test for an image classification task. For example, this pytorch tutorial uses RandomResizedCrop at train time and then Resize + ...
1
vote
1answer
78 views

Using MultiLabelBinarizer for SMOTE

This is my first NLP project. I'm trying to use SMOTE for a classifier with 14 classes. I need to convert the classes into an array before using SMOTE. I tried using MultiLinearBinarizer but it does ...
0
votes
1answer
58 views

ImageDataGenerator - trained with model.fit instead of model.fit_generator [closed]

I am a beginner in using the ImageDataGenerator from Keras and I accidentally used model.fit instead of model.fit_generator. ...
0
votes
0answers
35 views

How to build wake-word detection dataset from keyword pronunciations

I have quite a few audio files which are just pronunciation of one particular word, of course there are different tones, accents etc. I want to build model for wake-word detection, so I need to ...
1
vote
2answers
93 views

Same validation accuracy, different train accuracy for two neural networks models

I'm performing emotion classification over FER2013 dataset. I'm trying to measure different models performance, and when I checked ImageDataGenerator with a model I had already used I came up with the ...
1
vote
1answer
23 views

Data augmentation in deep training

I'm trying to understand the role of data augmentation and how it can affect the performance/accuracy of a deep model. My target application is a fire classification (fire or not, on video frames), ...
1
vote
0answers
18 views

How many augmentated data points for each training image?

What are some useful rules of thumb for picking the number of augmenters per training image? I realize this is a hyperparameter I can vary and test: I'm just trying to get a sense for reasonable ...
3
votes
1answer
749 views

Is GridSearchCV in combination with ImageDataGenerator possible and recommendable?

I want to optimize some hyperparameters for a CNN architecture by using GridSearchCV (Scikit-Learn) in combination with Data Augmentation (ImageDataGenerator from Keras). However, GridSearchCV only ...
1
vote
1answer
49 views

How important is the channel order in deep-learning computer vision tasks?

I stumbled across this question while working with OpenCV, which stores color images in BGR order in memory, while most other libraries I know of use RGB order. How important is this difference? ...
0
votes
0answers
28 views

Computer vision and Augmented reality

I'm new to the Computer Vision I want to make AR application, I don't know where to start. Please give me some advice?
1
vote
1answer
347 views

Fastest batch perspective transform for image augmentation

I need to do some augmentation for my training images for a neural networks. The problem is that even when loading batches in parallel, the augmentation is taking longer to perform than the network ...
0
votes
0answers
34 views

Random crop in multi-label image classification context

In some research papers, people use random cropping with variable sizes and then they resize them to original size as a data augmentation technique saying that it helps boost results. Can someone help ...
2
votes
0answers
350 views

Data Augmentation techniques for classification of imbalanced time series datasets

Now I have a task to classify the imbalanced time series datasets using ML classifiers, such as Logistic Regression, Decision Tree, SVM, and KNN. I am not allowed to use the Deep Learning tools, such ...
0
votes
0answers
41 views

How do I augment data after spliting traininng datset into train and validation set for CIFAR10 using PyTorch?

When classifying the CIFAR10 in PyTorch, there are normally 50,000 training samples and 10,000 testing samples. However, if I need to create a validation set, I can do it by splitting the training set ...
1
vote
1answer
25 views

What is the reason behind having low results using the data augmentation technique in NLP?

I used Data augmentation technique on my dataset, to have more data to train. My data is text so the data augmentation technique is based on random insertion of words, random swaps and synonyms ...
1
vote
1answer
318 views

Data Augmentation Multi Outputs

This question is asked several times here on SE, but I havent been able to find the right answer. I'm trying to build a network with 1 input and 2 outputs. I don't have a lot of data so I would like ...
1
vote
1answer
26 views

Using data agumentation for a frozen pre-trained model

I was following the following article with regards to doing transfer learning: https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html In the section, Using ...
0
votes
1answer
62 views

Non-Real Time Data Augmentation for CNN Classification. What are the drawbacks?

When people talk about and use data augmentation, are they mostly referring to real-time data augmentation? In the case of image classification, that would involve augmenting the data right before ...
2
votes
2answers
267 views

after augmentation validation accuracy going down?

My main question is about augmentation. if I process the augmentation I believe it always better than less data but in my case the validation accuracy going down train : 7000 images , validation: ...
3
votes
0answers
25 views

Back-Translation model for German and English

Do you know of any pre-trained models for back translation between German and English? I am aware that there are ways to include a monolingual corpus into the training of a machine translation model (...
2
votes
1answer
224 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 ...
2
votes
1answer
301 views

Is image sharpening a good idea for data augmentation?

I'm training segmentation networks and while the dataset is somehow decent (~5k images) I wanted to augment it, so far I'm trying: RandomFlip RandomRotate RandomBrightness changes RandomShadows Due ...
2
votes
1answer
302 views

Implementing back translation as a data augmentation for text classification

Since back translation English->other language -> English seems like quite a useful data augmentation technique , I wanted to experiment with it. E.g. it occurred to me that languages from very ...
2
votes
3answers
2k views

Is there any augmentation tool for images and bounding boxes?

I don't have a lot of training data and I'm looking for some tools in python or executable program like labelimg that do some heavy augmentation on images, even better if they also change bounding ...
7
votes
1answer
3k views

CNN - imbalanced classes, class weights vs data augmentation

I have a set of data with a few strongly imbalanced classes, eg the smallest class is about 54 times smaller than the largest. Therefore, data augmentation in order to equalize the size of classes ...
0
votes
1answer
673 views

How to properly rotate image and labels for semantic segmentation data augmentation in Tensorflow?

What's a proper procedure for doing the image and label rotation for semantic segmentation in dataset augmentation using Tensorflow? Images I have seen the function tf.contib.image.rotate(), but ...
1
vote
2answers
216 views

Is there a disadvantage to letting a model train for a large number of epochs?

I created a model to solve a time series forecasting problem. I had a limited amount of time series with which I could train the model therefore I decided to augment the data. The data augmentation ...
0
votes
1answer
23 views

Generated training set on convnet

I have a dataset with roughly 800 images that are classified in 18 classes. The classes are spread unevenly, with some classes having 30 images and others having 5. In order to increase my dataset,...