Preprocessing in scene text recognition - deep-learning

i'm new to OCR and has some questions about how to build an scene text recognition (STR) system.
Specifically saying in a STR dataset, the texts are various in lengths, fonts, and image sizes etc. Therefore, I suppose it's not a common sense to resize all the images to the same size before putting it inside a deep model for training. In this case, what's the right way to deal with this kind of data?

Related

Size of image for prediction with SageMaker object detection?

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.

how to format the image data for training/prediction when images are different in size?

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 .

achieve better recognition results via training tesseract

I have a question regarding achieving better recognition results with tesseract. I am using tesseract to recognize serial numbers. The serial numbes consist of only one font-type, characters A-Z, 0-9 and occur in different sizes and lengths.
At the moment I am able to recognize about 40% of the serial number images correct. Images are taken via mobile phone camera. Therefore the image quality isn't the best.
Special problem characters are 8/B, 5/6. Since I am recognizing only serial numbers, I am not using any dictionary improvements and every character is recognized independently.
My question is: Does someone has already experience in achieving better recognition results with training tesseract? How many images would be needed to be able to get good results.
For training tesseract should I use printed and afterwards photographed serial numbers, or should I use original digital serial numbers, without printing and photographing?
Maybe somebody has already experience in that kind of area.
Regarding training tesseract: I have already trained tesseract with some images. Therefore I have printed all characters in different sizes, photographed and labeled them correctly. Example training photo of the character 5
Is this a good/bad training example? Since I only want to recognize single characters without any dependency, I though I don't have to use words for training.
Actual I only have trained with 3 of these images for the characters B 8 6 5 which doesn't result in a better recognition in comparison with the original english (eng) tesseract database.
best regards,
Christoph
I am currently working on a Sikuli application using Tesseract to read text (Strings and numbers) from screenshots. I found that the best way to achieve accuracy was to process the screenshot before performing the OCR on it. However, most of the text I am reading is green text-on black background, making this my preferred solution. I used Scalr's method within BufferedImage to increase the size of the image:
BufferedImage bufImg = Scalr.resize(...)
which instantly yielded more accurate results with black text on gray background. I then used BufferedImage's options BufferedImage.TYPE_BYTE_GRAY and BufferedImage.TYPE_BYTE_BINARY when creating a new BufferedImage to process the Image to grayscale and black/white, respectively.
Following these steps brought Tesseract's accuracy from a 30% to around an 85% when dealing with green text on black background, and a really-close-to-100% accuracy when dealing with normal black text on white background. (sometimes letters within a word are mistaken by numbers i.e. hel10)
I hope this helps!

OCR algorithm- distinguish between textual image and object image

Am writing a program that extracts the contents from the logo of different websites.. i am using OCR to extract the text from the logo but i want to optimize the program and want to apply OCR only on those logos which have text but i dont know how to determine if a logo contains text or not??? any method??
this is a case where we need to know if an image has text in it. It is different from OCR.
The algorithm which is considered to be best to date is Stroke Width Transform. It was designed by Ephstein under Microsoft in 2010. It doesn't use any machine learning purposes.
You can get more details from this paper : Detecting Text in Natural Scenes with Stroke Width Transform
Or watch a video about this.
There is an implementation of this algorithm here.

document image processing

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.