Questions tagged [computer-vision]

For questions related to computer vision, which is an interdisciplinary scientific field (which can e.g. use image processing techniques) that deals with how computers can be made to gain high-level understanding from digital images or videos. For example, image recognition (that is, the identification of the type of objects in an image) is a computer vision problem.

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

Compute IoU for each class in Mask R-cnn

I'm trying to compute the IoU, with the matterport Mask R-cnn implementation, for each class (13 in total) that i have in my dataset. For now i managed to compute the average IoU for all the classes ...
1
vote
1answer
80 views

What is the current state of the art in animal facial recognition?

As far as I can tell, most work on facial recognition has been done in relation to human faces. Has any significant work been done for dogs, and are there any special challenges that would make it ...
1
vote
1answer
75 views

Questions about a research paper on salient region detection and segmentation

I am reading this paper in an attempt to recreate the salient region detection and segmentation model employed. I have the following questions pertaining to section 3 of the paper and I would highly ...
3
votes
1answer
80 views

In Computer Vision, what is the difference between a transformer and attention?

Having been studying computer vision for a while, I still cannot understand what the difference between a transformer and attention is?
2
votes
1answer
156 views

Can a fully convolutional network always return an image of the same size as the original?

I'm trying to perform a segmentation task on images of multiple sizes using fully convolutional neural networks. Currently, I'm using EfficientNet as a feature extractor, and adding a deconvolution/...
6
votes
1answer
58 views

Video summarization similar to Summe's TextRank

We have the popular TextRank API which given a text, ranks keywords and can apply summarization given a predefined text length. I am wondering if there is a similar tool for video summarization. ...
0
votes
1answer
47 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 ...
0
votes
0answers
16 views

Computing the mean attention distance for ViT

Recently I came across the paper that introduces the Vision Transformer (ViT) "AN IMAGE IS WORTH 16X16 WORDS: TRANSFORMERS FOR IMAGE RECOGNITION AT SCALE". The thing I don't really ...
1
vote
0answers
11 views

What is language-conditioned visual reasoning?

Can anyone explain what language-conditioned visual reasoning is? I saw this term in this paper and I searched on the internet but I couldn't find a proper explanation.
1
vote
0answers
14 views

What does the lambda parameter in the paper "Interpretable Explanations of Black Boxes by Meaningful Perturbation" do?

I do not understand the purpose of the $\lambda$ parameter in equation 3 of the paper Interpretable Explanations of Black Boxes by Meaningful Perturbation. $$m^{*}=\underset{m \in[0,1]^{\Lambda}}{\...
8
votes
1answer
231 views

Extending FaceNet’s triplet loss to object recognition

FaceNet uses a novel loss metric (triplet loss) to train a model to output embeddings (128-D from the paper), such that any two faces of the same identity will have a small Euclidean distance, and ...
0
votes
0answers
11 views

Save the Embeddings Model from a Complete Triplet Loss Network

I am working on a Siamese Neural Network with custom Triplet Loss function. As far as I learned from the documentations, we will train a complete network but we want to save only the CNN that is used ...
0
votes
0answers
16 views

How can I take continuous video input into my model?

Let's say I have designed an ML model that can take video input of a dog running around and give the breed of the dog as output. However, I do not want to wait for the video to finish before it is ...
0
votes
0answers
12 views

Comparing results of different image splicing methods on a part of the CASIA 2.0 dataset

So I am working on an image splicing detection algorithm using ResNet-50 model. I am using the CASIA 2.0 dataset which consists of 7491 Authentic images and 5123 Fake images. However out of the fake ...
1
vote
1answer
68 views

Are Markov Random Fields and Conditional Random Fields still used in computer vision?

Back before deep learning, there were a lot of different attempts at computer vision. Some involved Conditional Random Fields and Markov Random Fields, which were both computationally difficult and ...
0
votes
0answers
45 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 ...
0
votes
1answer
299 views

What is the state-of-the-art algorithm for neural style transfer?

I've read the paper A Neural Algorithm of Artistic Style by Gatys et. al. and I find the application of neural style transfer very fun. I also read that Exploring the structure of a real-time, ...
0
votes
0answers
21 views

Upscaling a low-res IR image with a high res webcam image

I have a low resolution thermal/IR image (for example 32x32 or 80x64) and a high resolution webcam image. I would like to combine the two to "fake" a high resolution thermal image (I can ...
3
votes
3answers
182 views

Does a fully convolutional network share the same translation invariance properties we get from networks that use max-pooling?

Does a fully convolutional network share the same translation invariance properties we get from networks that use max-pooling? If not, why do they perform as well as networks which use max-pooling?
0
votes
0answers
11 views

How to scale Computer Vision? How to implement Emotion detection from live video feed of N different video simultaneously?

I have a pipeline based on Scaled Yolov4 detection algorithm for faces which extract faces of users and then uses a CNN to ...
2
votes
2answers
66 views

Why adversarial images are not the mainstream for captchas?

In order to check, whether the visitor of the page is a human, and not an AI many web pages and applications have a checking procedure, known as CAPTCHA. These tasks are intended to be simple for ...
1
vote
3answers
173 views

Why is AI Super Resolution Reconstruction more than just guessing?

I saw a video on Youtube about AI and Super Resolution Image Reconstruction with TecoGAN. I must say I am impressed. Now, I am wondering how reliable this is. I have learned at university that you ...
1
vote
1answer
36 views

Why is this variable in equation 2 of the SQAIR paper a random vector of $n$ ones followed by a zero?

I've been reading the SQAIR paper lately, and the mathematics involved seems a bit complicated. Some background, about the paper: SQAIR stands for Sequential Attend, Infer, Repeat - the paper does ...
0
votes
0answers
26 views

Why the collection of background/negative image dataset is not taught in object detection tutorials and books?

While I was doing an object detection project, I have encountered the problem of getting FALSE POSITIVES and FALSE NEGATIVES. After days of research on StackOverflow, I figured out that I need to ...
0
votes
1answer
119 views

What is the difference between exhaustive nearest neighbor search and k-nearest neighbour search?

I have two lists of feature vectors calculated from pre-trained CNN for image retrieval task: Query: FV_Q and Reference FV_R. <...
0
votes
0answers
12 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
0answers
26 views

How to change number of trained layers in object detection TensorFlow models?

Training custom object detection models with TensorFlow usually means a transfer learning of pre-trained models and, if I understand it correctly, it means only training the few last layers, with ...
1
vote
0answers
33 views

What do "spatial" and "temporal" mean in the context of image processing?

I am new to image processing. I am trying to understand CNNs from this blog post. Here's an excerpt from that article that mentions these terms. A ConvNet is able to successfully capture the Spatial ...
1
vote
0answers
23 views

Why class embedding token is added to the Visual Transformer?

In the famous work on the Visual Transformers, the image is split into patches of a certain size (say 16x16), and these patches are treated as tokens in the NLP tasks. In order to perform ...
0
votes
0answers
11 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 ...
0
votes
0answers
35 views

Bounding box transformation using point cloud

I am working on a project in which I have an RGB Image and the corresponding 2d bounding boxes for the objects in the image. I also have a respective point cloud and therefore, I can extract the depth ...
0
votes
1answer
22 views

Changing a CNN-LSTM image captioning architecture to use BiLSTMs

Currently I'm dealing with an assignment that made us implement the network mentioned in this paper. The network has an architecture similar to this: As you can see it uses a Unidirectional RNN (in ...
1
vote
4answers
156 views

How much statistics is involved in AI?

This may be a stupid question but I couldn't really find an answer on the internet. I am a 3rd year math major who is interested in computer science, particularly algorithms and competitive ...
0
votes
0answers
21 views

How to remember previously detected objects in computer vision?

Let's say I have a drone that has to fly (and scan) over some area (blue on the images - in this case a PV power plant, but it could be anything) in an autonomous way like in the pic below: Let's ...
0
votes
0answers
10 views

How to measure(classify) the speed of oncoming traffic via Computer Vision and Neural Networks?

Suppose I have different videos of the same car sometimes moving slow, sometimes moving fast, say, at 50Kmph as slow and 60Kmph as fast. (Assume the background is a green screen and the car doesn't ...
2
votes
1answer
56 views

Is it possible to use self-supervised learning on different images for the pretext and downstream tasks?

I have just come across the idea of self-supervised learning. It seems that it is possible to get higher accuracies on downstream tasks when the network is trained on pretext tasks. Suppose that I ...
1
vote
0answers
19 views

What are the most relevant resources that define the face detection problem formally?

I am new to AI, and I am a bit lost about finding the relevant materials that define the face detection problem formally/mathematically. Can anyone help me formally define face detection, or at least ...
3
votes
1answer
80 views

What is a heatmap in the CornerNet paper?

I have been working on understanding how CornerNet works, but I couldn't figure out a few parts about the architecture. First, the authors mention that there are 3 distinct parts to be predicted as a ...
0
votes
1answer
56 views

How to use 'Canny/Watershed' algorithm's output as an input for Image Classification Model

I have a very silly problem in hand. I have implemented 2 methods which give me the mask to separate the objects from the background. What I get from one method is the object encapsulated in the red ...
0
votes
0answers
22 views

Arcface implementation for image similarity produces opposite embeddings for positive negative image pairs

So I've built an arcface model with this arcface layer implementation: https://github.com/4uiiurz1/keras-arcface/blob/master/metrics.py I trained for a few epochs and now that I'm comparing the ...
10
votes
2answers
1k views

Do deep learning algorithms represent ensemble-based methods?

According to the Wikipedia article on deep learning: Deep learning is a branch of machine learning based on a set of algorithms that attempt to model high-level abstractions in data by using a deep ...
2
votes
1answer
42 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 ...
0
votes
2answers
291 views

Unet Overfitting for binary segmentation of fake images

I am working on a project where I am trying to detect and localize forgeries in images. I am using the CASIA v2 dataset and using Unet model for the task. I have the binary masks of all the images in ...
0
votes
0answers
11 views

how to use Laplacian of mesh structure (LBO) for meshes that are registered in deep learning methods based on spectrum (ChebNet for isntance)?

In graph neural network frameworks, there is always a template with a shared structure among all graphs. I have meshes that are registered but obviously, Lalpalcian and their geometry are different. ...
0
votes
0answers
21 views

Is it possible to do object detection on an object classification dataset?

I'm new to computer vision, which I find fascinating. I wonder whether it is possible or if there has been any research into going from object recognition data to object detection. In other words, ...
0
votes
0answers
26 views

Which layer to route out of layers of same width and height in Yolo implementation?

In Yolo configuration files (like yolo3.cfg in dark-net), there are many layers with output of same height and width due to ...
0
votes
0answers
11 views

Can I get some advices on inferencing people from upwards using Yolov5?

I'm trying to inference people from upwards and count them using Yolov5. I know the controversy between yolov5 and yolov4, but for me, Yolov5 is more easier and reliable to use, also the setup. I have ...
0
votes
0answers
11 views

How to obtain part filter anchors in DPM Detector

The DPM detector (https://cs.brown.edu/people/pfelzens/papers/lsvm-pami.pdf) uses latent-svm to train the weights of the root and part filters. During training for positive samples, it alternates ...
0
votes
0answers
11 views

Should I L-2 Normalise outputs in Siamese Neural Neural Network for distance computation for Triplet Loss or not?

I am building a Siamese Neural Network for Images (CNN) which uses the FaceNet's Triplet Loss as its loss function. I found a good Implementation here where we build a model and the outputs from the ...
2
votes
0answers
30 views

How is the data labelled in order to train a region proposal network?

I don't get how the training of the RPN works. From the forward propagation, I have $W \times H \times k$ outputs from the RPN. How is the training data labeled such that I can use the loss function ...

1
2
3 4 5
9