Questions tagged [image-classification]

For questions about image classification: a decision problem where an algorithm must decide to which class ('cat', 'chair', 'tree') an input image belongs.

281 questions with no upvoted or accepted answers
Filter by
Sorted by
Tagged with
4
votes
2answers
2k views

Pytorch - Loss is decreasing but Accuracy not improving

It seems loss is decreasing and the algorithm works fine. But accuracy doesn't improve and stuck. ...
4
votes
1answer
919 views

What preprocessing steps to be followed before image comparison?

1 down vote favorite For example I am trying to find the similarity between two images using skimage - SSIM. The code block will be as follows ...
3
votes
0answers
87 views

ResNet50 + Transformer

In many papers people extract features from image using ResNet and than pass them through transformer. I want to implement the same. I want to get features and than classify them using transformer. ...
3
votes
1answer
60 views

Can I train a CNN to detect the number of objects without localizing them first?

So I was trying to search but couldn't find any answers. I was wondering if it possible to train a model to detect the number of items of interest in a photo without having bounding boxes or dots to ...
3
votes
1answer
940 views

Weighted loss functions vs weighted sampling?

For image classification tasks, is there a practical difference between using weighted loss functions vs. using weighted sampling? (I would appreciate theoretical arguments, experience or published ...
3
votes
0answers
77 views

What happens with activations?

I am playing with convolution network, assembling something between AlexNet and ResNet. Not very deep, about 10 conv. layers including 2 through residual connection, and 3 fully-connected layes at the ...
3
votes
1answer
249 views

Why is training and validation loss steadily rising (eventually to NaN) in this CNN of mine?

Dear ML and data scientists: I have 4 layers of gray scale images for every single biological specimen in my dataset. I am trying to train a 4-convolution CNN (see pytorch architecture below) to ...
3
votes
2answers
560 views

Pre-processing on MRI images

I have MRI images of brain tumors collected from a hospital (not a benchmark dataset). And I am planning to use them to predict/classify tumour types using a typical machine learning approach: texture ...
3
votes
0answers
57 views

How to help neuronal network with an other model

I am working on an image classification problem, the input data normally is images to classify, but I thought latitude and longitude would play something on these satellite images. I sorted by ...
3
votes
3answers
72 views

Is it reliable to use TensorFlow (ML in general) to classify baggage bag tags based on the presence of a green stripe?

The images are identical except for the presence of the stripe on the side. I am trying to use a classify the images into 2 classes: greenStripe, noGreenStripe. I tried to use tensorflow retrain with ...
3
votes
0answers
318 views

Spatial Transformer Networks and Data Augmentation

We are all familiar with the famous Deep Mind paper STN. Upon implementation, such as here, did anyone still use input data augementation such as affine transformations? There are used to make CNN ...
3
votes
0answers
5k views

Adding multilabel classifier to TensorFlow example

I am starting with the generic TensorFlow example. To classify my data I need to use multiple labels (ideally multiple softmax classifiers) on the final layer, ...
3
votes
1answer
1k views

Generate new images based on a dataset of images

Is it possible take a training set of one million 120 x 120 pixel tiles, feed these tiles into a machine learning algorithm. And then make it synthesize images that look like the original training set ...
2
votes
1answer
37 views

Transfer Deep Learning from one aerial imagery datset to many others

I am new to Deep Learning but have been able to use RasterVision successfully to predict building footprints within a set of aerial imagery. This aerial imagery data set is for a province of New ...
2
votes
0answers
18 views

How will my CNN results be affected by large discrepancies between the number of samples in some of the classes?

The number of samples in my dataset range between 3800 and 100,000 per class. Was wondering if my neural network will be more biased towards the classes with a higher number of images. I'm trying on ...
2
votes
0answers
172 views

TensorFlow - TFRecords load and transform images with bounding boxes

I'm trying to build a 'Car Classifier' using TensorFlow. I have 1000 labelled JPG images, 800x800, complete with bounding boxes and associated annotations.coco.json; split into train/validate/test ...
2
votes
1answer
216 views

Training CNN on a huge data set

I am trying to train an AlexNet image model on the RVL-CDIP Dataset. The dataset consists of 320,000 training images, 40,000 validation images, and 40,000 test images. Since the dataset is huge I ...
2
votes
1answer
20 views

features to help distinguish between document images

we are trying to build a model to classify different types of documents as the first step in our pipeline (final goal is to read all the text). Currently we use ImageNet to extract the features and ...
2
votes
0answers
104 views

Understanding the significance of LeNet-5 w/ MNIST data set

I'm beginning to learn about conv nets and started with what I understand to be one of the seminal works: LeNet-5. However, my limited experimentation doesn't seem to show any advantage over a single ...
2
votes
0answers
25 views

Classifying satellite data

I have a large data set of RGB satellite data that classifies 64x64 pixel images with a spatial resolution of 10m per pixel into 10 classes (e.g. highway, industrial, river, forest). Now I want to ...
2
votes
1answer
14 views

Distinguish real building pictures from rendered ones

I need to train an algorithm to identify rendered building images from real pictures. Here's an example: A rendered picture A real picture The real pictures will vary from Good quality to Poor ...
2
votes
3answers
186 views

Can the same CNN architecture be used for different data sets?

I have a CNN architecture that works well on 32x32x3 images. Can I use that same architecture for a data set made up of 28x28x1 images? (Both data sets have 10 classes). If this is possible, what ...
2
votes
3answers
43 views

Classifier design for website screenshots

I'm working on a project that requires determining if the page representing a hosted file on a third party platform (such as rapidgator or nitroflare) is still up or not. For example, here is a file ...
2
votes
0answers
26 views

List of CNN for Emotion/Sentiment recognition on images with performance on main datasets (IAPS, GAPED, EmoPics, NAPS)

There are more and more databases of pictures classified or rated with emotions. For instance, I know of 4 databases (IAPS, GAPED, EmoPics, NAPS) rating pictures on 2 dimensions: Valance (positive vs ...
2
votes
2answers
67 views

Does high accuracy metrics with small (but equally sampled) dataset means a good model?

I have been training my CNN with 200 images per class for a classification problem. There problem is a binary classification one. And with the amount of test data ( 25 per class) I am getting good ...
2
votes
0answers
23 views

Does object detection do a better job at image classification than image classification

I read in an article that object segmentation can do object detection better than object detection algorithms. I assume this is because there is more detailed information in the annotation images. I ...
2
votes
2answers
803 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: ...
2
votes
0answers
605 views

Why is my Siamese network always predicting 1?

There is no change to loss and the accuracy stays the same. ...
2
votes
0answers
199 views

Audio files and their corresponding spectrograms for image classification process

Suppose I have a dataset of audio files that I have to use for whale sound classification. I am choosing the strategy of treating it as an image classification problem by using their corresponding ...
2
votes
0answers
18 views

Where should we release our video dataset for activity recognition?

We have an activity recognition dataset made of tens of thousands video clips. Which are the alternatives to release it to the research community? Is there any public website? I saw for instance ...
2
votes
0answers
20 views

Neural Network for detecting/checking for requirements in diagrams

My question is more about what approach is a good/the best approach for my problem: THE PROBLEM - I'm an (mechanical/software) engineer and we take extensive amount of time to review technical ...
2
votes
0answers
40 views

How can I get the original pixels that lead to the decision in CNN?

I work on medical images and 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 and get the features maps. Now I ...
2
votes
0answers
117 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 ...
2
votes
0answers
52 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 ...
2
votes
0answers
31 views

How to turn classification tree results into a GIS map

I'm new-ish to machine learning, so this could be a silly question. Apologies if so. The idea is is to predict groundwater occurrence based on a regression tree. This is my conceptual model: Target ...
2
votes
0answers
33 views

How does convolution operation perform in CNN?

I know that convolution operation has the property, associative law. if I need to use multiple filters in succession and to perform this operation on multiple images, it makes sense to convolve the ...
2
votes
0answers
190 views

Machine Learning & Image Recognition: How to start?

I've been a full stack web developer for 15 years now and would like to be involved in machine learning. There is already a specific scenario for this: We have a database with several million products ...
2
votes
1answer
136 views

calculation of average ROC in IMageNet paper?

The IMageNEt paper Image Net. presents the Average ROC curve for the 16 classes in the imagenet data, visit image figure. 8 in the paper. what is the known function to compute this ROC plot. As ROC ...
2
votes
1answer
544 views

Two parallel models for semantic segmentation in Keras

I want to build two parallel models for image semantic segmentation in Keras. ...
2
votes
0answers
1k views

application of Histogram of oriented gradients in colored image

I recently learned about face recognition with deep learning here. One of the approach involved is Histogram of oriented gradients which is used for face detection as follows (short summary) : ...
2
votes
0answers
131 views

What are natural (computed) pre-images useful for?

I've just been reading Zeiler, M.D. and Fergus, R., 2014, September. Visualizing and understanding convolutional networks. In European Conference on Computer Vision (pp. 818-833). Springer ...
2
votes
1answer
1k views

Test data from ImageNet

Here's the description about the data usage for ILSVRC 2016 of ImageNet. I've interpreted it with the table as follows, ...
2
votes
0answers
17 views

Number of SURF features normally used?

I am interesting in knowing an approximate amount of SURF features used to do e.g. image-recognition on 224x224x3 images. Is there a standard amount of features used? Or perhaps a well-acclaimed ...
2
votes
0answers
1k views

Caffe net.predict() outputs random results (GoogleNet)

I used pretrained GoogleNet from https://github.com/BVLC/caffe/tree/master/models/bvlc_googlenet and finetuned it with my own data (~ 100k images, 101 classes). After one day training I achieved 62% ...
1
vote
0answers
22 views

Neural network training for two vs. many classes

Suppose I want to train a convolutional neural network to distinguish not between dogs and cats, but between images of dogs and images of any feline – cat, lion, tiger, leopard, cheetah, etc. ...
1
vote
0answers
14 views

how to train image classification with small difference in low-level feature

I would like to build the model that capable of doing classification for example 2 classes below : I tried alexnet, resnet50, resnet18, vgg16 but seem they are failed to differentiate between this ...
1
vote
1answer
37 views

How can I classify multiple images "simultaneously"?

I am dealing with a multiclass image classification problem with N classes. Particularly interesting is now that a single instance is NOT a single image as you would expect normally, but rather a set ...
1
vote
0answers
22 views

Which model is used for document extraction (CamScanner, Microsoft Lens etc)

I want to start a small project where I'd create a model(s) that would extract document from a picture and rescale it, something like CamScanner or Microsoft Lens apps do. I've gathered a small ...
1
vote
1answer
32 views

spot/stain growth in image classification problems

I am working on a problem with images where we are monitoring development of spot in certain region of image. We are able to classify spot present(NOK) or not present(OK) successfully if initially ...
1
vote
0answers
11 views

Image Classification: Different Background on Training and Test Set

I am currently working on a challenge for a course at my university. We need to assign the image of a leaf to one of 14 different classes and we are given an unbalanced training set of ~17k pictures (...

1
2 3 4 5 6