I'm looking for U-net implementation for landmark detection task, where the architecture is intended to be similar to the figure above. For reference please see this: An Attention-Guided Deep Regression Model for Landmark Detection in Cephalograms
From the figure, we can see the input dimension is 572x572 but the output dimension is 388x388. My question is, how do we visualize and correctly understand the cropped output? From what I know, we ideally expect the output size is the same as input size (which is 572x572) so we can apply the mask to the original image to carry out segmentation. However, from some tutorial like (this one), the author recreate the model from scratch then use "same padding" to overcome my question, but I would prefer not to use same padding to achieve same output size.
I couldn't use same padding because I choose to use pretrained ResNet34 as my encoder backbone, from PyTorch pretrained ResNet34 implementation they didn't use same padding on the encoder part, which means the result is exactly similar as what you see in the figure above (intermediate feature maps are cropped before being copied). If I would to continue building the decoder this way, the output will have smaller size compared to input image.
The question being, if I want to use the output segmentation maps, should I pad its outside until its dimension match the input, or I just resize the map? I'm worrying the first one will lost information about the boundary of image and also the latter will dilate the landmarks predictions. Is there a best practice about this?
The reason I must use a pretrained network is because my dataset is small (only 100 images), so I want to make sure the encoder can generate good enough feature maps from the experiences gained from ImageNet.
After some thinking and testing of my program, I found that PyTorch's pretrained ResNet34 didn't loose the size of image because of convolution, instead its implementation is indeed using same padding. An illustration is
Input(3,512,512)-> Layer1(64,128,128) -> Layer2(128,64,64) -> Layer3(256,32,32)
-> Layer4(512,16,16)
so we can use deconvolution (or ConvTranspose2d in PyTorch) to bring the dimension back to 128, then dilate the result 4 times bigger to get the segmentation mask (or landmarks heatmaps).
Related
I was going through vggnet paper and i came across the testing phase of vggnet.
During the testing phase, test image goes through the vggnet and a class score map is obtained. This class score map is spatially averaged to produce a fixed size vector.
I have googled class score map, but then i couldn't find any relevant results. I wish to know what is the role of class score map.
Any hint would be greatly helpful. Thanks
When you train an image recognition model, you train it for a specific image size (and resolution), let's say n_dims = [256, 256]. Now, in the prediction phase, you have images of different sizes (with respect to pixels), e.g. [1024, 1024]. You extract patches (you can resize the image first by lowering the resolution) and hover over the image patches with your model, and for each patch, you obtain a prediction for all classes (in a patch, more than one of the objects might be present), which you have to average somehow for the whole image at the end.
See OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks.
Instead, we explore the entire image by densely running the network at each location and at multiple
scales. While the sliding window approach may be computationally prohibitive for certain types
of model, it is inherently efficient in the case of ConvNets (see section 3.5). This approach yields
significantly more views for voting, which increases robustness while remaining efficient. The result
of convolving a ConvNet on an image of arbitrary size is a spatial map of C-dimensional vectors at
each scale.
I'm using the AWS SageMaker "built in" object detection algorithm (SSD) and we've trained it on a series of annotated 512x512 images (image_shape=512). We've deployed an endpoint and when using it for prediction we're getting mixed results.
If the image we use for prediciton is around that 512x512 size we're getting great accuracy and good results. If the image is significantly larger (e.g. 8000x10000) we get either wildly inaccurate, or no results. If I manually resize those large images to 512x512pixels the features we're looking for are no longer discernable to the eye. Which suggests that if my endpoint is resizing images, then that would explain why the model is struggling.
Note: Although the size in pexels is large, my images are basically line drawings on a white background. They have very little color and large patches of solid white, so they compress very well. I'm mot running into the 6Mb request size limit.
So, my questions are:
Does training the model at image_shape=512 mean my prediction images should also be that same size?
Is there a generally accepted method for doing object detection on very large images? I can envisage how I might chop the image into smaller tiles then feed each tile to my model, but if there's something "out of the box" that will do it for me, then that'd save some effort.
Your understanding is correct. The endpoint resizes images based on the parameter image_shape. To answer your questions:
As long as the scale of objects (i.e., expansion of pixels) in the resized images are similar between training and prediction data, the trained model should work.
Cropping is one option. Another method is to train separate models for large and small images as David suggested.
Suppose I have a CNN which is trained for classifying images of different animals, the output of such model will be a point (output point) in a n spatial dimension, where n is the number of animal classes the model is trained on; then that output is transformed in a way to convert it into a one-hot vector of n parameters, giving then the correct label for the image from the point of view of the CNN, but let's stick with the n dimensional point, which is the concept of an input image.
Suppose then that I want to take that point and transform it in a way so that the final output is an image with constraint width and height (the dimensions should be the same with different input images) which outputs the same point as the input image's, how do I do that?
I'm basically asking for the methods used (training mostly) for this kind of task, where an image must be reconstructed based on the output point of the CNN -I know the image will never be identical, but I'm looking for images that generate the same (or at least not so different) output point as a input image when that point is inputted to the CNN-. Take in mind that the input of the model I'm asking for is n and the output is a two (or three if it's not in grayscale) dimensional tensor. I noticed that deepdream does exactly this kind of thing (I think), but every time I put "deepdream" and "generate" in Google, an online generator is almost always shown, not the actual techniques; so if there are some answers to this I'd love to hear about them.
The output label does not contain enough information to reconstruct an entire image.
Quoting from the DeepDream example ipython notebook:
Making the "dream" images is very simple. Essentially it is just a gradient ascent process that tries to maximize the L2 norm of activations of a particular DNN layer.
So the algorithm modifies an existing image such that outputs of certain nodes in the network (can be in an intermediate layer, not necessarily the output nodes) become large. In order to do that, it has to calculate the gradient of the node output with respect to the input pixels.
I am trying to train my model which classifies images.
The problem I have is, they have different sizes. how should i format my images/or model architecture ?
You didn't say what architecture you're talking about. Since you said you want to classify images, I'm assuming it's a partly convolutional, partly fully connected network like AlexNet, GoogLeNet, etc. In general, the answer to your question depends on the network type you are working with.
If, for example, your network only contains convolutional units - that is to say, does not contain fully connected layers - it can be invariant to the input image's size. Such a network could process the input images and in turn return another image ("convolutional all the way"); you would have to make sure that the output matches what you expect, since you have to determine the loss in some way, of course.
If you are using fully connected units though, you're up for trouble: Here you have a fixed number of learned weights your network has to work with, so varying inputs would require a varying number of weights - and that's not possible.
If that is your problem, here's some things you can do:
Don't care about squashing the images. A network might learn to make sense of the content anyway; does scale and perspective mean anything to the content anyway?
Center-crop the images to a specific size. If you fear you're losing data, do multiple crops and use these to augment your input data, so that the original image will be split into N different images of correct size.
Pad the images with a solid color to a squared size, then resize.
Do a combination of that.
The padding option might introduce an additional error source to the network's prediction, as the network might (read: likely will) be biased to images that contain such a padded border.
If you need some ideas, have a look at the Images section of the TensorFlow documentation, there's pieces like resize_image_with_crop_or_pad that take away the bigger work.
As for just don't caring about squashing, here's a piece of the preprocessing pipeline of the famous Inception network:
# This resizing operation may distort the images because the aspect
# ratio is not respected. We select a resize method in a round robin
# fashion based on the thread number.
# Note that ResizeMethod contains 4 enumerated resizing methods.
# We select only 1 case for fast_mode bilinear.
num_resize_cases = 1 if fast_mode else 4
distorted_image = apply_with_random_selector(
distorted_image,
lambda x, method: tf.image.resize_images(x, [height, width], method=method),
num_cases=num_resize_cases)
They're totally aware of it and do it anyway.
Depending on how far you want or need to go, there actually is a paper here called Spatial Pyramid Pooling in Deep Convolution Networks for Visual Recognition that handles inputs of arbitrary sizes by processing them in a very special way.
Try making a spatial pyramid pooling layer. Then put it after your last convolution layer so that the FC layers always get constant dimensional vectors as input . During training , train the images from the entire dataset using a particular image size for one epoch . Then for the next epoch , switch to a different image size and continue training .
I working on an application for processing document images (mainly invoices) and basically, I'd like to convert certain regions of interest into an XML-structure and then classify the document based on that data. Currently I am using ImageJ for analyzing the document image and Asprise/tesseract for OCR.
Now I am looking for something to make developing easier. Specifically, I am looking for something to automatically deskew a document image and analyze the document structure (e.g. converting an image into a quadtree structure for easier processing). Although I prefer Java and ImageJ I am interested in any libraries/code/papers regardless of the programming language it's written in.
While the system I am working on should as far as possible process data automatically, the user should oversee the results and, if necessary, correct the classification suggested by the system. Therefore I am interested in using machine learning techniques to achieve more reliable results. When similar documents are processed, e.g. invoices of a specific company, its structure is usually the same. When the user has previously corrected data of documents from a company, these corrections should be considered in the future. I have only limited knowledge of machine learning techniques and would like to know how I could realize my idea.
The following prototype in Mathematica finds the coordinates of blocks of text and performs OCR within each block. You may need to adapt the parameters values to fit the dimensions of your actual images. I do not address the machine learning part of the question; perhaps you would not even need it for this application.
Import the picture, create a binary mask for the printed parts, and enlarge these parts using an horizontal closing (dilation and erosion).
Query for each blob's orientation, cluster the orientations, and determine the overall rotation by averaging the orientations of the largest cluster.
Use the previous angle to straighten the image. At this time OCR is possible, but you would lose the spatial information for the blocks of text, which will make the post-processing much more difficult than it needs to be. Instead, find blobs of text by horizontal closing.
For each connected component, query for the bounding box position and the centroid position. Use the bounding box positions to extract the corresponding image patch and perform OCR on the patch.
At this point, you have a list of strings and their spatial positions. That's not XML yet, but it sounds like a good starting point to be tailored straightforwardly to your needs.
This is the code. Again, the parameters (structuring elements) of the morphological functions may need to change, based on the scale of your actual images; also, if the invoice is too tilted, you may need to "rotate" roughly the structuring elements in order to still achieve good "un-skewing."
img = ColorConvert[Import#"http://www.team-bhp.com/forum/attachments/test-drives-initial-ownership-reports/490952d1296308008-laura-tsi-initial-ownership-experience-img023.jpg", "Grayscale"];
b = ColorNegate#Binarize[img];
mask = Closing[b, BoxMatrix[{2, 20}]]
orientations = ComponentMeasurements[mask, "Orientation"];
angles = FindClusters#orientations[[All, 2]]
\[Theta] = Mean[angles[[1]]]
straight = ColorNegate#Binarize[ImageRotate[img, \[Pi] - \[Theta], Background -> 1]]
TextRecognize[straight]
boxes = Closing[straight, BoxMatrix[{1, 20}]]
comp = MorphologicalComponents[boxes];
measurements = ComponentMeasurements[{comp, straight}, {"BoundingBox", "Centroid"}];
texts = TextRecognize#ImageTrim[straight, #] & /# measurements[[All, 2, 1]];
Cases[Thread[measurements[[All, 2, 2]] -> texts], (_ -> t_) /; StringLength[t] > 0] // TableForm
The paper we use for skew angle detection is: Skew detection and text line position determination in digitized documents by Gatos et. al. The only limitation with this paper is that it can detect skew upto -5 and +5 degrees. After that, we need something to slap the user with a message! :)
In your case, where there are primarily invoice scans, you may beautifully use: Multiresolution Analysis in Extraction of Reference Lines from Documents with Gray Level Background by Tag et. al.
We wrote the code in MATLAB, if you need help let me know!
I worked on a similar project once, and for being a long time user of OpenCV I ended up using it once again. OpenCV is a popular-cross-platform-computer-vision-library that offers programming interfaces for C and C++.
I found an interesting blog that had a post on how to detect the skew angle of a text using OpenCV, and then another on how to deskew.
To retrieve the text of the document and be able to pass a smaller image to tesseract, I suggest taking a look at the bounding box technique.
I don't know if the image acquisition procedure is your responsibility, but if it is you might want to take a look at how to do camera calibration with OpenCV to fix the distortion in the image caused by some camera lenses.