请选择 进入手机版 | 继续访问电脑版

技术控

    今日:187| 主题:53563
收藏本版 (1)
最新软件应用技术尽在掌握

[其他] Automatically grading multiple choice exams from photos using Python and OpenCV

[复制链接]
你穿过的墙 发表于 2016-10-3 22:00:03
256 2

立即注册CoLaBug.com会员,免费获得投稿人的专业资料,享用更多功能,玩转个人品牌!

您需要 登录 才可以下载或查看,没有帐号?立即注册

x

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
  
  Over the past few months I’ve gotten quite the number of requests landing in my inbox to build a bubble sheet/Scantron-like test reader using computer vision and image processing techniques.
  And while I’ve been having a lot of fun doing this series on machine learning and deep learning, I’d be    lyingif I said this little mini-project wasn’t a short, welcome break. One of my favorite parts of running the PyImageSearch blog is demonstrating how to build     actualsolutions to problems using computer vision.  
  In fact, what makes this project    so specialis that we are going to combine the techniques from     manyprevious blog posts, including    building a document scanner,contour sorting, and    perspective transforms. Using the knowledge gained from these previous posts, we’ll be able to make quick work of this bubble sheet scanner and test grader.  
  You see, last Friday afternoon I quickly Photoshopped an example bubble test paper, printed out a few copies,    and then set to work on coding up the actual implementation.  
  Overall, I am quite pleased with this implementation and I think you’ll absolutely be able to use this bubble sheet grader/OMR system as a starting point for your own projects.
  To learn more about utilizing computer vision, image processing, and OpenCV to automatically grade bubble test sheets,     keep reading.  
      Looking for the source code to this post?
Jump right to the downloads section.    Bubble sheet scanner and test grader using OMR, Python, and OpenCV

  In the remainder of this blog post, I’ll discuss what exactly    Optical Mark Recognition(OMR) is. I’ll then demonstrate how to implement a bubble sheet test scanner and grader using     strictlycomputer vision and image processing techniques, along with the OpenCV library.  
  Once we have our OMR system implemented, I’ll provide sample results of our test grader on a few example exams, including ones that were filled out with nefarious intent.
  Finally, I’ll discuss some of the shortcomings of this current bubble sheet scanner system and how we can improve it in future iterations.
  What is Optical Mark Recognition (OMR)?

  Optical Mark Recognition, or OMR for short, is the process of    automaticallyanalyzing human-marked documents and interpreting their results.  
  Arguably, the most famous, easily recognizable form of OMR are          bubble sheet multiple choice tests    , not unlike the ones you took in elementary school, middle school, or even high school.  
  If you’re unfamiliar with “bubble sheet tests” or the trademark/corporate name of “Scantron tests”, they are simply multiple-choice tests that you take as a student. Each question on the exam is a multiple choice — and you use a #2 pencil to mark the “bubble” that corresponds to the correct answer.
  The most notable bubble sheet test you experienced (at least in the United States) were taking the SATs during high school, prior to filling out college admission applications.
  I    believethat the SATs use the software provided by Scantron to perform OMR and grade student exams, but I could easily be wrong there. I only make note of this because Scantron is used in over 98% of all US school districts.  
  In short, what I’m trying to say is that there is a    massive marketfor Optical Mark Recognition and the ability to grade and interpret human-marked forms and exams.  
  Implementing a bubble sheet scanner and grader using OMR, Python, and OpenCV

  Now that we understand the basics of OMR, let’s build a computer vision system using Python and OpenCV that can    readand     gradebubble sheet tests.  
  Of course, I’ll be providing lots of visual example images along the way so you can understand    exactly what techniques I’m applyingand     why I’m using them.  
  Below I have included an example filled in bubble sheet exam that I have put together for this project:
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 1:The example, filled in bubble sheet we are going to use when developing our test scanner software.   
    We’ll be using this as our example image as we work through the steps of building our test grader. Later in this lesson, you’ll also find additional sample exams.
  I have also included a    blank exam templateas a .PSD (Photoshop) file so you can modify it as you see fit. You can use the           “Downloads”    section at the bottom of this post to download the code, example images, and template file.  
  The 7 steps to build a bubble sheet scanner and grader

  The goal of this blog post is to build a bubble sheet scanner and test grader using Python and OpenCV.
  To accomplish this, our implementation will need to satisfy the following 7 steps:
  
       
  •       Step #1:Detect the exam in an image.   
  •       Step #2:Apply a perspective transform to extract the top-down, birds-eye-view of the exam.   
  •       Step #3:Extract the set of bubbles (i.e., the possible answer choices) from the perspective transformed exam.   
  •       Step #4:Sort the questions/bubbles into rows.   
  •       Step #5:Determine the marked (i.e., “bubbled in”) answer for each row.   
  •       Step #6:Lookup the correct answer in our answer key to determine if the user was correct in their choice.   
  •       Step #7:Repeat for all questions in the exam.  
  The next section of this tutorial will cover the actual    implementationof our algorithm.  
  The bubble sheet scanner implementation with Python and OpenCV

  To get started, open up a new file, name it                  test_grader        .        py          , and let’s get to work:  
  1. # import the necessary packages
  2. from imutils.perspectiveimport four_point_transform
  3. from imutilsimport contours
  4. import numpyas np
  5. import argparse
  6. import imutils
  7. import cv2
  8.  
  9. # construct the argument parse and parse the arguments
  10. ap = argparse.ArgumentParser()
  11. ap.add_argument("-i", "--image", required=True,
  12. help="path to the input image")
  13. args = vars(ap.parse_args())
  14.  
  15. # define the answer key which maps the question number
  16. # to the correct answer
  17. ANSWER_KEY = {0: 1, 1: 4, 2: 0, 3: 3, 4: 1}
复制代码
On    Lines 2-7we import our required Python packages.  
  You should already have OpenCV and Numpy installed on your system, but you    mightnot have the most recent version of    imutils, my set of convenience functions to make performing basic image processing operations easier. To install                  imutils          (or upgrade to the latest version), just execute the following command:  
  1. $ pipinstall --upgradeimutils
复制代码
Lines 10-12parse our command line arguments. We only need a single switch here,                  --        image          , which is the path to the input bubble sheet test image that we are going to grade for correctness.  
  Line 17then defines our                  ANSWER_KEY          .  
  As the name of the variable suggests, the                  ANSWER_KEY          provides integer mappings of the     question numbersto the     index of the correct bubble.  
  In this case, a key of    0indicates the     first question, while a value of     1signifies     “B”as the correct answer (since     “B”is the index     1in the string     “ABCDE”). As a second example, consider a key of     1that maps to a value of     4— this would indicate that the answer to the second question is     “E”.  
  As a matter of convenience, I have written the entire answer key in plain english here:
  
       
  •       Question #1:B   
  •       Question #2: E   
  •       Question #3: A   
  •       Question #4: D   
  •       Question #5: B  
  Next, let’s preprocess our input image:
  1. # load the image, convert it to grayscale, blur it
  2. # slightly, then find edges
  3. image = cv2.imread(args["image"])
  4. gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
  5. blurred = cv2.GaussianBlur(gray, (5, 5), 0)
  6. edged = cv2.Canny(blurred, 75, 200)
复制代码
On    Line 21 we load our image from disk, followed by converting it to grayscale (    Line 22), and blurring it to reduce high frequency noise (    Line 23).  
  We then apply the Canny edge detector on    Line 24to find the     edges/outlinesof the exam.  
  Below I have included a screenshot of our exam after applying edge detection:
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 2:Applying edge detection to our exam neatly reveals the outlines of the paper.   
    Notice how the edges of the document are    clearly defined, with     all four vertices of the exambeing present in the image.  
  Obtaining this silhouette of the document is    extremely importantin our next step as we will use it as a marker to apply a perspective transform to the exam, obtaining a top-down, birds-eye-view of the document:  
  1. # find contours in the edge map, then initialize
  2. # the contour that corresponds to the document
  3. cnts = cv2.findContours(edged.copy(), cv2.RETR_EXTERNAL,
  4. cv2.CHAIN_APPROX_SIMPLE)
  5. cnts = cnts[0] if imutils.is_cv2() else cnts[1]
  6. docCnt = None
  7.  
  8. # ensure that at least one contour was found
  9. if len(cnts) > 0:
  10. # sort the contours according to their size in
  11. # descending order
  12. cnts = sorted(cnts, key=cv2.contourArea, reverse=True)
  13.  
  14. # loop over the sorted contours
  15. for c in cnts:
  16. # approximate the contour
  17. peri = cv2.arcLength(c, True)
  18. approx = cv2.approxPolyDP(c, 0.02 * peri, True)
  19.  
  20. # if our approximated contour has four points,
  21. # then we can assume we have found the paper
  22. if len(approx) == 4:
  23. docCnt = approx
  24. break
复制代码
Now that we have the outline of our exam, we apply the                  cv2        .        findContours          function to find the lines that correspond to the exam itself.  
  We do this by sorting our contours by their    area(from largest to smallest) on     Line 37(after making sure at least one contour was found on     Line 34, of course). This implies that larger contours will be placed at the front of the list, while smaller contours will appear farther back in the list.  
  We make the assumption that our exam will be the    main focal point of the image, and thus be larger than other objects in the image. This assumption allows us to “filter” our contours, simply by investigating their area and knowing that the contour that corresponds to the exam should be near the front of the list.  
  However,    contour area and sizeis not enough — we should also check the number of     verticeson the contour.  
  To do, this, we loop over each of our (sorted) contours on    Line 40. For each of them, we approximate the contour, which in essence means we     simplifythe number of points in the contour, making it a “more basic” geometric shape. You can read more about contour approximation in this post on    building a mobile document scanner.  
  On    Line 47we make a check to see if our approximated contour has four points, and if it does,     we assume that we have found the exam.  
  Below I have included an example image that demonstrates the                  docCnt          variable being drawn on the original image:  
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 3:An example of drawing the contour associated with the exam on our original image, indicating that we have successfully found the exam.   
    Sure enough, this area corresponds to the outline of the exam.
  Now that we have used contours to find the outline of the exam, we can apply a perspective transform to obtain a top-down, birds-eye-view of the document:
  1. # apply a four point perspective transform to both the
  2. # original image and grayscale image to obtain a top-down
  3. # birds eye view of the paper
  4. paper = four_point_transform(image, docCnt.reshape(4, 2))
  5. warped = four_point_transform(gray, docCnt.reshape(4, 2))
复制代码
In this case, we’ll be using my implementation of the                  four_point_transform          function which:  
  
       
  • Orders the       (x, y)-coordinates of our contours in a       specific, reproducible manner.   
  • Applies a perspective transform to the region.  
  You can learn more about the perspective transform inthis postas well as    this updated one on coordinate ordering, but for the time being, simply understand that this function handles taking the “skewed” exam and transforms it, returning a top-down view of the document:  
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 4:Obtaining a top-down, birds-eye view of both the original image      (left)along with the grayscale version      (right).   
    Alright, so now we’re getting somewhere.
  We found our exam in the original image.
  We applied a perspective transform to obtain a 90 degree viewing angle of the document.
  But how do we go about actually    gradingthe document?  
  This step starts with    binarization, or the process of thresholding/segmenting the     foregroundfrom the     backgroundof the image:  
  1. # apply Otsu's thresholding method to binarize the warped
  2. # piece of paper
  3. thresh = cv2.threshold(warped, 0, 255,
  4. cv2.THRESH_BINARY_INV | cv2.THRESH_OTSU)[1]
复制代码
After applying Otsu’s thresholding method, our exam is now a    binaryimage:  
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 5:Using Otsu’s thresholding allows us to segment the foreground from the background of the image.   
    Notice how the    background of the image is     black, while the     foregroundis     white.  
  This binarization will allow us to once again apply contour extraction techniques to find each of the bubbles in the exam:
  1. # find contours in the thresholded image, then initialize
  2. # the list of contours that correspond to questions
  3. cnts = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL,
  4. cv2.CHAIN_APPROX_SIMPLE)
  5. cnts = cnts[0] if imutils.is_cv2() else cnts[1]
  6. questionCnts = []
  7.  
  8. # loop over the contours
  9. for c in cnts:
  10. # compute the bounding box of the contour, then use the
  11. # bounding box to derive the aspect ratio
  12. (x, y, w, h) = cv2.boundingRect(c)
  13. ar = w / float(h)
  14.  
  15. # in order to label the contour as a question, region
  16. # should be sufficiently wide, sufficiently tall, and
  17. # have an aspect ratio approximately equal to 1
  18. if w >= 20 and h >= 20 and ar >= 0.9 and ar <= 1.1:
  19. questionCnts.append(c)
复制代码
Lines 64-67handle finding contours on our                  thresh          binary image, followed by initializing                  questionCnts          , a list of contours that correspond to the questions/bubbles on the exam.  
  To determine which regions of the image are bubbles, we first loop over each of the individual contours (    Line 70).  
  For each of these contours, we compute the bounding box (    Line 73), which also allows us to compute the    aspect ratio, or more simply, the ratio of the width to the height (    Line 74).  
  In order for a contour area to be considered a bubble, the region should:
  
       
  • Be sufficiently wide and tall (in this case, at least 20 pixels in both dimensions).   
  • Have an aspect ratio that is      approximatelyequal to 1.  
  As long as these checks hold, we can update our                  questionCnts          list and mark the region as a bubble.  
  Below I have included a screenshot that has drawn the output of                  questionCnts          on our image:  
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 6:Using contour filtering allows us to find all the question bubbles in our bubble sheet exam recognition software.   
    Notice how    onlythe question regions of the exam are highlighted and nothing else.  
  We can now move on to the “grading” portion of our OMR system:
  1. # sort the question contours top-to-bottom, then initialize
  2. # the total number of correct answers
  3. questionCnts = contours.sort_contours(questionCnts,
  4. method="top-to-bottom")[0]
  5. correct = 0
  6.  
  7. # each question has 5 possible answers, to loop over the
  8. # question in batches of 5
  9. for (q, i) in enumerate(np.arange(0, len(questionCnts), 5)):
  10. # sort the contours for the current question from
  11. # left to right, then initialize the index of the
  12. # bubbled answer
  13. cnts = contours.sort_contours(questionCnts[i:i + 5])[0]
  14. bubbled = None
复制代码
First, we must sort our                  questionCnts          from top-to-bottom. This will ensure that rows of questions that are     closer to the topof the exam will appear    firstin the sorted list.  
  We also initialize a bookkeeper variable to keep track of the number of                  correct          answers.  
  On    Line 90we start looping over our questions. Since each question has 5 possible answers, we’ll apply NumPy array slicing and contour sorting to to sort the     current set of contoursfrom left to right.  
  The reason this methodology works is because we have    alreadysorted our contours from top-to-bottom. We     knowthat the 5 bubbles for each question will appear sequentially in our list —     but we       do not know whether these bubbles will be sorted from left-to-right.     The sort contour call on     Line 94takes care of this issue and ensures each row of contours are sorted into rows, from left-to-right.  
  To visualize this concept, I have included a screenshot below that depicts each row of questions as a separate color:
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 7:By sorting our contours from top-to-bottom, followed by left-to-right, we can extract each row of bubbles. Therefore, each row is equal to the bubbles for one question.   
    Given a row of bubbles, the next step is to determine which bubble is filled in.
  We can accomplish this by using our                  thresh          image and counting the number of non-zero pixels (i.e.,     foreground pixels) in each bubble region:  
  1. # loop over the sorted contours
  2. for (j, c) in enumerate(cnts):
  3. # construct a mask that reveals only the current
  4. # "bubble" for the question
  5. mask = np.zeros(thresh.shape, dtype="uint8")
  6. cv2.drawContours(mask, [c], -1, 255, -1)
  7.  
  8. # apply the mask to the thresholded image, then
  9. # count the number of non-zero pixels in the
  10. # bubble area
  11. mask = cv2.bitwise_and(thresh, thresh, mask=mask)
  12. total = cv2.countNonZero(mask)
  13.  
  14. # if the current total has a larger number of total
  15. # non-zero pixels, then we are examining the currently
  16. # bubbled-in answer
  17. if bubbledis None or total > bubbled[0]:
  18. bubbled = (total, j)
复制代码
Line 98handles looping over each of the sorted bubbles in the row.  
  We then construct a mask for the current bubble on    Line 101and then count the number of non-zero pixels in the masked region (    Lines 107 and 108). The more non-zero pixels we count, then the more foreground pixels there are, and therefore the bubble with the maximum non-zero count is the index of the bubble that the the test taker has bubbled in (    Line 113 and 114).  
  Below I have included an example of creating and applying a mask to each bubble associated with a question:
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 8:An example of constructing a mask for each bubble in a row.   
    Clearly, the bubble associated with    “B”has the most thresholded pixels, and is therefore the bubble that the user has marked on their exam.  
  This next code block handles looking up the correct answer in the                  ANSWER_KEY          , updating any relevant bookkeeper variables, and finally drawing the marked bubble on our image:  
  1. # initialize the contour color and the index of the
  2. # *correct* answer
  3. color = (0, 0, 255)
  4. k = ANSWER_KEY[q]
  5.  
  6. # check to see if the bubbled answer is correct
  7. if k == bubbled[1]:
  8. color = (0, 255, 0)
  9. correct += 1
  10.  
  11. # draw the outline of the correct answer on the test
  12. cv2.drawContours(paper, [cnts[k]], -1, color, 3)
复制代码
Based on whether the test taker was correct or incorrect yields which color is drawn on the exam. If the test taker is    correct, we’ll highlight their answer in     green. However, if the test taker made a mistake and marked an incorrect answer, we’ll let them know by highlighting the     correctanswer in     red:  
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 9:Drawing a “green” circle to mark “correct” or a “red” circle to mark “incorrect”.   
    Finally, our last code block handles scoring the exam and displaying the results to our screen:
  1. $ pipinstall --upgradeimutils
  2. 0
复制代码
Below you can see the output of our fully graded example image:
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 10:Finishing our OMR system for grading human-taken exams.   
    In this case, the reader obtained an 80% on the exam. The only question they missed was #4 where they incorrectly marked    “C”as the correct answer (    “D”was the correct choice).  
  Why not use circle detection?

  After going through this tutorial, you might be wondering:
  “Hey Adrian, an answer bubble is a circle. So why did you extract contours instead of applyingHough circles to find the circles in the image?”  
  Great question.
  To start, tuning the parameters to Hough circles on an image-to-image basis can be a real pain. But that’s only a minor reason.
  The    real reasonis:  
  User error.
  How many times, whether purposely or not, have you filled in outside the lines on your bubble sheet? I’m not expert, but I’d have to guess that at least 1 in every 20 marks a test taker fills in is “slightly” outside the lines.
  And guess what?
  Hough circles don’t handle deformations in their outlines very well — your circle detection would totally fail in that case.
  Because of this, I instead recommend using contours and contour properties to help you filter the bubbles and answers. The                  cv2        .        findContours          function doesn’t care if the bubble is “round”, “perfectly round”, or “oh my god, what the hell is that?”.  
  Instead, the                  cv2        .        findContours          function will return a set of     blobsto you, which will be the foreground regions in your image. You can then take these regions process and filter them to find your questions (as we did in this tutorial), and go about your way.  
  Our bubble sheet test scanner and grader results

  To see our bubble sheet test grader in action, be sure to download the source code and example images to this post using the          “Downloads”    section at the bottom of the tutorial.  
  We’ve already seen                  test_01        .        png          as our example earlier in this post, so let’s try                  test_02        .        png          :  
  1. $ pipinstall --upgradeimutils
  2. 1
复制代码
Here we can see that a particularly nefarious user took our exam. They were not happy with the test, writing    “#yourtestsux”across the front of it along with an anarchy inspiring     “#breakthesystem”. They also marked     “A”for all answers.  
  Perhaps it comes as no surprise that the user scored a pitiful 20% on the exam, based entirely on luck:
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 11:By using contour filtering, we are able to ignore the regions of the exam that would have otherwise compromised its integrity.   
    Let’s try another image:
  1. $ pipinstall --upgradeimutils
  2. 2
复制代码
This time the reader did a little better, scoring a 60%:
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 12:Building a bubble sheet scanner and test grader using Python and OpenCV.   
    In this particular example, the reader simply marked all answers along a diagonal:
  1. $ pipinstall --upgradeimutils
  2. 3
复制代码
     

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 13:Optical Mark Recognition for test scoring using Python and OpenCV.   
    Unfortunately for the test taker, this strategy didn’t pay off very well.
  Let’s look at one final example:
  1. $ pipinstall --upgradeimutils
  2. 4
复制代码
         Figure 14:Recognizing bubble sheet exams using computer vision.   
    This student clearly studied ahead of time, earning a perfect 100% on the exam.
  Extending the OMR and test scanner

  Admittedly, this past summer/early autumn has been one of the    busiestperiods of my life, so I needed to    timeboxthe development of the OMR and test scanner software into a single, shortened afternoon last Friday.  
  While I was able to get the barebones of a    workingbubble sheet test scanner implemented, there are certainly a few areas that need improvement. The most obvious area for improvement is the     logic to handle non-filled in bubbles.  
  In the current implementation, we (naively) assume that a reader has filled in    oneand     only onebubble per question row.  
  However, since we determine if a particular bubble is “filled in” simply by counting the number of thresholded pixels in a row and then sorting in descending order, this can lead to two problems:
  
       
  • What happens if a user       does not bubble in an answer for a particular question?   
  • What if the user is nefarious and marks       multiplebubbles as “correct” in the same row?  
  Luckily, detecting and handling of these issues isn’t terribly challenging, we just need to insert a bit of logic.
  For issue #1, if a reader chooses    notto bubble in an answer for a particular row, then we can place a     minimum thresholdon     Line 108where we compute                  cv2        .        countNonZero          :  
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 15:Detecting if a user has marked zero bubbles on the exam.   
    If this value is sufficiently large, then we can mark the bubble as “filled in”. Conversely, if                  total          is too small, then we can skip that particular bubble. If at the end of the row there are     no bubbles with sufficiently large threshold counts, we can mark the question as “skipped” by the test taker.  
  A similar set of steps can be applied to issue #2, where a user marks    multiplebubbles as correct for a single question:  
      

Automatically grading multiple choice exams from photos using Python and OpenCV

Automatically grading multiple choice exams from photos using Python and OpenCV
    Figure 16:Detecting if a user has marked multiple bubbles for a given question.   
    Again, all we need to do is apply our thresholding and count step, this time keeping track if there are    multiple bubblesthat have a                  total          that exceeds some pre-defined value. If so, we can invalidate the question and mark the question as incorrect.  
  Summary

  In this blog post, I demonstrated how to build a bubble sheet scanner and test grader using computer vision and image processing techniques.
  Specifically, we implemented    Optical Mark Recognition (OMR) methods that facilitated our ability of capturing human-marked documents and     automaticallyanalyzing the results.  
  Finally, I provided a Python and OpenCV implementation that you can use for building your own bubble sheet test grading systems.
  If you have any questions, please feel free to leave a comment in the comments section!
  But before you, be sure to enter your email address in the form below to be notified when future tutorials are published on the PyImageSearch blog!
  Downloads:
更多图片 小图 大图
组图打开中,请稍候......



上一篇:watchOS by Tutorials Updated for Swift 3 and watchOS 3
下一篇:Animate Your React Native App
weeks_less 发表于 2016-10-4 02:35:30
我小学十年,中学十二年,我被评为全校最熟悉的面孔,新老师来了都跟我打听学校内幕……
回复 支持 反对

使用道具 举报

心想事成1 发表于 2016-10-5 18:52:04
你穿过的墙很是无聊啊
回复 支持 反对

使用道具 举报

*滑动验证:
您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

我要投稿

推荐阅读

扫码访问 @iTTTTT瑞翔 的微博
回页顶回复上一篇下一篇回列表
手机版/CoLaBug.com ( 粤ICP备05003221号 | 文网文[2010]257号 )

© 2001-2017 Comsenz Inc. Design: Dean. DiscuzFans.

返回顶部 返回列表