<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Vasugupta on Medium]]></title>
        <description><![CDATA[Stories by Vasugupta on Medium]]></description>
        <link>https://medium.com/@vasugupta2000?source=rss-2ff227372f60------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*N43YCF9cUK5RKl8Dxfh9lw.jpeg</url>
            <title>Stories by Vasugupta on Medium</title>
            <link>https://medium.com/@vasugupta2000?source=rss-2ff227372f60------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 11:28:52 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@vasugupta2000/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Role of AI for Disabled People]]></title>
            <link>https://vasugupta2000.medium.com/role-of-ai-for-disabled-people-b387becf95f3?source=rss-2ff227372f60------2</link>
            <guid isPermaLink="false">https://medium.com/p/b387becf95f3</guid>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[disabled]]></category>
            <category><![CDATA[smart-home]]></category>
            <dc:creator><![CDATA[Vasugupta]]></dc:creator>
            <pubDate>Thu, 11 Nov 2021 11:36:35 GMT</pubDate>
            <atom:updated>2021-11-11T11:38:12.599Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*045ro4sqhYGrTum78V3WJQ.jpeg" /><figcaption>AI Improving accessibility</figcaption></figure><h3><strong>What’s AI?</strong></h3><p>Artificial Intelligence is also known as AI, is the simulation of human intelligence because it can be programmed to mimic human actions. Thus, we called it the artificial human brain. In today’s world AI is playing a major role in our lives, starting from recognizing the sleep and voice pattern to recommending us the choice of food we eat. AI has been proven the boon to today’s era having the big potential to automate the task performed by the human such as speech and voice recognition, sentiment analysis, pose estimation, face recognition and detection and so many more.</p><p>So advancement in AI has proven to be a game-changer for people having some disability by making it easier to create some interactive tools.<br> This blog will go through some useful examples of AI which helped in improving the lives of disabled people.</p><h3><strong>Applications of AI in the lives of disabled people</strong></h3><ol><li><strong>Human Connection and Communication</strong></li></ol><p>The AI voice assistant technology like Alexa, Google Assistant, Cortana, Siri are the perfect example we have today that has created a new means of accessibility for disabled people. They can access all the daily general information just by using their voice and speaking near the mic of their mobile devices. For example, they can ask the above-mentioned technologies the virtual assistants to set the alarms, to narrate the latest news, can listen to any song by just saying “Play the music”, can perform arithmetic operations, language conversions, and many more that a human can do.</p><p>The speech-to-text and text-to-speech technologies helped those people who had speech impediments. These technologies learn from the speaker’s pronunciation over time and translate it into the text or in the form of audio so that they can communicate face to face with normal people. Also, for the people who are deaf and dumb, AI also has developed a solution to cater to their problems. The technology of translation of Sign language to text/speech and vice-versa too has made the life of disabled people much easier. Now they can communicate with the outer world smoothly. There is an AI app of google called Deepmind which used the capabilities of AI to develop a lip-reading algorithm that can detect the speech of an individual and then convert it into the voice/text.</p><p>Recently an AI product was developed which uses the technology of computer vision and created a life-changing opportunity for people with low vision. The product is Microsoft seeing AI that narrates the entire world around them visually impaired by reading texts, describing whereabouts and the looks of the nearby people by identifying and recognizing faces and emotions.</p><p><strong>2. Independent in Home Living</strong></p><p>AI is also used in creating smart home technology that can be of great benefit for handicapped people or people with low mobility. For example, by some simple and easy voice commands they can control electric items of their home which are IoT enabled, can request the information they are seeking thus helping the disabled people live an independent life.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/318/1*KAqE3-oZd2sVfvIGqcZxIg.jpeg" /><figcaption>Smart Devices</figcaption></figure><p>For example, Google Echo dot and Amazon Alexa both are very useful because both can interact with the IoT-enabled smart home devices and thus help people with disabilities by allowing voice interaction, setting reminders and alarms, access to real-time information, etc. One can literally control the lightning of their use from their smartphone either by an app or by voice commands. Smart Doorbell and Smart garage openers have made life easy and handy for disabled people. These virtual assistants too can be used as a friend in the alone time. They can talk like humans and answer the questions asked by the person. Thus, making the life of disabled people little happening.</p><p><strong>3. Increasing Accessibility</strong></p><p>Autonomous vehicles are in trend and their success is due to AI technology. These vehicles can be beneficial to people living with limited physical mobility. Thanks to the companies who came up with this idea and developed a product out of it namely Google, Tesla, Uber, Lyft, Drive AI, Here Technologies, etc. Using these autonomous driverless vehicles, they can go outside for work, grocery shopping, etc and they won’t be dependent on the other for this.</p><p>To conclude I would say that AI-powered products and devices have totally changed the life of every individual and they are able to make a great difference in the lives of disabled people by supporting them in daily activities and has opened many opportunities for them to see this world from a new angle. The advancement in AI is still in progress and we may see some new technologies in the coming future.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b387becf95f3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Implementation of Gradient Ascent using Logistic Regression]]></title>
            <link>https://vasugupta2000.medium.com/implementation-of-gradient-ascent-using-logistic-regression-7f5343877c21?source=rss-2ff227372f60------2</link>
            <guid isPermaLink="false">https://medium.com/p/7f5343877c21</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[gradient]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Vasugupta]]></dc:creator>
            <pubDate>Sun, 01 Nov 2020 19:48:21 GMT</pubDate>
            <atom:updated>2020-11-10T08:31:26.129Z</atom:updated>
            <content:encoded><![CDATA[<p>Before understanding the gradient ascent let’s first understand what is Logistic Regression?</p><h3><strong>Logistic Regression</strong></h3><p>Logistic Regression is the machine learning classification algorithm which is used in predictive analysis. Logistic regression is almost similar to Linear regression but the main difference here is the cost function. Logistic Regression uses much more complex function namely log-likelihood Cost function whereas the other uses mean squared error(MSE) as the cost function. This function is based on the concept of probability and for a single training input (x,y), the assumption made by the function is</p><figure><img alt="Mathematical Assumption" src="https://cdn-images-1.medium.com/max/293/1*S71AhoGzCbD0z-h4gbDn_g.png" /></figure><p>These probability values are then mapped by hypothesis which is <strong>Sigmoid function </strong>between 0 and 1.</p><h3><strong>What is Sigmoid Function?</strong></h3><p>It is used to map predictions to the range of 0 and 1. Basically the value that is received from performing hypothesis is passed into this function which maps it between 0 and 1. So it outputs the probability value.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/843/1*IsR40Cc6BfY9bLtR4c6euQ.png" /><figcaption>Sigmoid Function</figcaption></figure><p><strong>So lets look at the implementation of the sigmoid function.</strong></p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/011c5f6b8591ae03894088b5a9d6f533/href">https://medium.com/media/011c5f6b8591ae03894088b5a9d6f533/href</a></iframe><p>In the above Source code, function takes a single parameter as a input that is the numpy array(z) and then returns the numpy array of mapped probability value between 0 and 1. The self in the function just represent the instance of the class. It has nothing to do with the output of the function.</p><h4>Let’s talk about the cost function</h4><p>Cost function is basically the loss function in terms of linear regression but in Gradient ascent we need to maximize the cost function.</p><p>So in Machine Learning, the cost functions are calculated as how far is the predicted value from the actual value. Thus the difference between them is either minimized for Gradient descent or is Maximized for Gradient Ascent. In Gradient ascent it is called as <strong>Log Likelihood Estimation or Maximum Likelihood estimation. </strong>Let’s see how we can perform log likelihood estimation.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/90066e5c5c83dd824caf220896058a23/href">https://medium.com/media/90066e5c5c83dd824caf220896058a23/href</a></iframe><p>So since its a Cost Function so we have to pass the actual value and the predicted value as the parameter which is nothing but numpy array here.</p><p>So we compute the log likelihood here which is calculated by the given formula.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/843/1*-Yj38Tg-zh58jO1fBPUJow.png" /><figcaption>Log Likelihood</figcaption></figure><p>So while calculating the Gradient(slope) in the Gradient ascent which will be discussed further in this blog, we take the partial differentiation of the above function with respect to (theta) to find maximum likelihood.</p><p>You might be thinking that numpy array above may have a value 0 and at 0 Log(0) is undefined, then to prevent this condition we can take some minimum value instead of 0 to avoid undefine condition.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/915/1*HQ-Rajh95u4oUT3eq7zOEg.png" /></figure><p>So here if 0 comes then it takes the minimum value which is assigned to the variable ‘self.eps’.</p><h3>Gradient Ascent</h3><p>So after seeing the cost function and sigmoid function, now let’s think of an algorithm which can combine these two function and gives us the desired result. The algorithm is the <strong>Gradient Ascent</strong> algorithm.</p><p>So Gradient Ascent is an iterative optimization algorithm for finding local maxima of a differentiable function. The algorithm moves in the direction of gradient calculated at each and every point of the cost function curve till the stopping criteria meets.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/0*PfkFznKzQ0te2YUF.png" /><figcaption>Gradient Ascent</figcaption></figure><p>Now we have the log likelihood function, so we simply need to choose the values of θ that maximizes the function. In short we will partially derivate the function with respect to θ(theta). Here θ is the weight matrix which can be assigned to 0 or to any random values between 0 and 1 in the starting.</p><p>So lets derivate the function.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/624/1*S1tQUk3BIJWl5Yug_uI6nA.png" /></figure><p>so we got the last term as the partial derivative of the log_likelhood</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/292/1*S3jjeULRlz805xIGFPYY7w.png" /><figcaption>Partial Derivative of Log Likelihood</figcaption></figure><p>Now we have partial derivative, so our goal is to choose the parameters (θ) that maximizes the likelihood. Here comes the Gradient Ascent. The idea behind gradient ascent is that gradient points ‘uphill’. So if you slowly slowly moves towards the direction of gradient then you eventually make it to the global maxima. Gradient ascent has an analogy in which we have to imagine ourselves at the bottom of a mountain valley and left stranded and blindfolded, our objective is to reach the top of the hill.</p><p>Now to maximize our log likelihood we need to run the gradient ascent function on each parameter i.e.</p><blockquote><strong><em>weights = weights + learning_rate*gradient</em></strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/283/1*uvi4wUl_9M0MEj80tJVhPg.png" /><figcaption>Gradient Ascent Function</figcaption></figure><p>Here α is the Learning rate of the model, which is the step size that we take towards uphill. So by running this function we reach to the optimum θ where the Likelihood is maximum.</p><p><strong>Lets look at the code of Gradient Ascent</strong></p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/985b0b0996e416dcf49b7c0f6fc7f1e3/href">https://medium.com/media/985b0b0996e416dcf49b7c0f6fc7f1e3/href</a></iframe><p>So in the above function we take X (X_train) and y (y_train) as input which are numpy ndarray. First we initialise the weights (θ) matrix with 0’s or any random value between 0 and 1.</p><p>Now we perform hypothesis and calculate the probability values of the input data ‘X’. Then we calculate the gradient of the log_likelihood function which is just the partial derivative of the function with respect to weights. Then the weights are updated using gradient ascent function. This process is repeated for N number of iterations specified by the user. So slowly we reach to the maxima with the help of gradient ascent.</p><p>This is basically the fitting of the model and now lets evaluate our model if its correctly predicting the result or not.</p><p>For this lets take some dataset. Download the dataset as done below. I am using the divorce dataset. You can use any other classification dataset.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/973767a9f3a68a5c593a2b2378c03bd1/href">https://medium.com/media/973767a9f3a68a5c593a2b2378c03bd1/href</a></iframe><p>Perform Pre-processing like Label/One Hot Encoding, normalize it using MinMax Scaler, add a bias, check for missing values if yes then input it. Here the dataset which I have chosen is already label encoded and has short range of values so their is no need of normalizing. Otherwise normalization is must.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/f64a87760b535c454eaf444c84b432b9/href">https://medium.com/media/f64a87760b535c454eaf444c84b432b9/href</a></iframe><p>Now we have the data frame so lets segregate the original data into training and testing data. For segregation we will use <strong>train_test_split</strong> function of sklearn library. It is advisable to split the dataset in the ratio of (70:30, 75:25, 80:20).</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/13e0cccf27ca3e6fa8d5676ecfee8104/href">https://medium.com/media/13e0cccf27ca3e6fa8d5676ecfee8104/href</a></iframe><p>In this particular code we take all the rows and n-1 columns of the dataframe in ‘x’ as the input &amp; all the rows and last column in the ‘y’ as the target. Further the dataframe is converted into numpy array.</p><p>Now you have X_train and y_train, so first initialize the model and pass these two into the fit function in the gradient ascent created above. Here default learning rate is taken as 0.01 which is the advisable to start with. Though in the initial phases of learning do try to play with the hyperparameters to study its affect on the model.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/a72440c3d0aa01bf99416de06d6ce28c/href">https://medium.com/media/a72440c3d0aa01bf99416de06d6ce28c/href</a></iframe><p>Now the model will be trained once you run the fit function. Now comes the prediction, all these above tasks are being performed to make prediction on the new sample of data. So let’s so the prediction on the testing set of the data that was segregated above. Simultaneously we will create the function to compute the accuracy of the model on testing set.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/91619aedd09528143a5cdff5d8d83a88/href">https://medium.com/media/91619aedd09528143a5cdff5d8d83a88/href</a></iframe><p>So after this whole hard work of understanding algorithm and writing the code we finally got the accuracy which was<strong> 88.46153846153846</strong>.</p><p>We may even plot the graph of Log likelihood vs No. of Iterations and it result into the perfect smooth curve as expected.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/5b5565c9b756625b7b774e3283e00fed/href">https://medium.com/media/5b5565c9b756625b7b774e3283e00fed/href</a></iframe><figure><img alt="" src="https://cdn-images-1.medium.com/max/527/1*Qo0Q42jkS5vtdbOEdQSN0w.png" /><figcaption>Log Likelihood Curve</figcaption></figure><p>Thus i will sum up the whole code of the model for your understanding. Try the implementation with different types of classification datasets, and do let me in the comments if you face any difficulty.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/ce8efa405971c3941b2823e0a88da9d4/href">https://medium.com/media/ce8efa405971c3941b2823e0a88da9d4/href</a></iframe><p>In the above predict function, i passed the numpy array as the input which is the list of sample whose values we need to classify. The values of the array are first passed to the predict_proba() function which return the probability values. These values are then converted into the 1 or 0 depending upon the threshold value set by us. Here the value of threshold is set to be as 0.5, then if the value returned from predict_proba is greater than equal to 0.5 it is scaled to 1 otherwise downscaled to 0.</p><h3><strong>Conclusion</strong></h3><p>In this blog, i have presented you with the basic concept of the <strong>Gradient Ascent Algorithm</strong> with the example. Hope you have understood its implementation and is highly motivated towards Artificial Intelligence and Machine Learning.</p><p><strong>Always remember that Artificial Intelligence is the New Electricity and my friend you are the lineman producing and controlling it.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7f5343877c21" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[EXTRACTING INFORMATION FROM MEDICINE USING DEEP LEARNING AND COMPUTER VISION]]></title>
            <link>https://vasugupta2000.medium.com/extracting-information-from-medicine-using-deep-learning-and-computer-vision-84a5b6f2dad4?source=rss-2ff227372f60------2</link>
            <guid isPermaLink="false">https://medium.com/p/84a5b6f2dad4</guid>
            <category><![CDATA[blog]]></category>
            <category><![CDATA[computer-vision]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[speech-recognition]]></category>
            <category><![CDATA[visually]]></category>
            <dc:creator><![CDATA[Vasugupta]]></dc:creator>
            <pubDate>Sun, 05 Jul 2020 09:55:21 GMT</pubDate>
            <atom:updated>2020-07-07T19:25:22.911Z</atom:updated>
            <content:encoded><![CDATA[<p>In this world, where technology brings the world to our fingertips, Deep Learning plays a vital role in solving the major issues in today’s world.</p><p>So What is Deep Learning?</p><p>Deep Learning is basically the subset of the machine learning or Artificial Intelligence which has the neural networks that are capable of learning unsupervised from unstructured or unlabeled data</p><h3>Introduction to the Problem</h3><p>Visually impaired people suffer a lot in society. Being disable is either by the mercy of God or by any external factor that led to this. With all this disability they even face many day to day problems. One of them is being fooled by others be it in the supermarkets or by the auto rickshaw’s driver etc. Even they are fooled by shrewd chemists. No one focuses on this problem but it exists in our society. To stop people from getting cheated, our vision is to design a tool that can update the off-the-shelf inventory by just scanning the images of medicines.</p><figure><img alt="Visually Impaired Person using stick to walk." src="https://cdn-images-1.medium.com/max/304/1*VafW-3P2ByuF8tdnJINWhg.png" /><figcaption>Visually Impaired Person using stick to walk.</figcaption></figure><p>The proposed solution is used to extract some basic but important details from that package of medicine like name of that medicine, manufacturing date, expiry date, MRP(maximum retail price) of that medicine and convert them to speech format.</p><p>We propose a model to detect<strong> </strong>and recognize the text from the images and convert them into speech using a deep learning framework.The process of extracting text data in a machine/computer readable format from real world images is quite a complex and challenging task in the computer vision community..</p><p>The ultimate goal is to design an end-to-end framework, which internally has two steps: Text Detection and Recognition.</p><p>This problem mainly has three important functionalities:</p><ol><li>Text Detection</li><li>Text Recognition</li><li>Converting text into speech</li></ol><h3><strong>Dataset Used</strong></h3><ul><li>The Dataset used to train the model was the dataset provided to the <strong>ICDAR 2019</strong>. Robust Reading Challenge on Multi-Lingual scene Text detection and Recognition.</li><li>The Training set is composed of <strong>10,000 images </strong>with<strong> 10,000 text files </strong>having the coordinates of the word by word of the text in the image with the text of the image<strong>.</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jPuyrCC5citqxY7W" /></figure><h3><strong>EAST (Efficient Accurate Scene Text detector)</strong></h3><p>This is a very effective method for text detection. It is only a proper text detection method. It can find horizontal and rotated bounding boxes. It can be combined with any text recognition method. EAST can detect text both in images and in the video. So we re-implemented EAST Text detector using TensorFlow to develop the model of our own need.</p><h3><strong>Methodology</strong></h3><p>ResNet-50 is an already preexisting Convolutional Neural Network that is 50 layers deep. So we retrained that neural network on our dataset. The training was done at 1000 epochs and then extracted the frozen model of it with the accuracy of 81%.</p><p>Below is the training sample.</p><figure><img alt="Training Sample" src="https://cdn-images-1.medium.com/max/920/0*FxE5Gq6bw4d3iFgw" /><figcaption>Training Sample</figcaption></figure><p>The trained model was then used to detect and localize the text on the image.</p><p>The image was then cropped according to the rectangular box<strong> </strong>created on it. Then this image was passed as a parameter into the pytesseract<strong> </strong>library to extract the text. This text is then converted into the speech using pyttsx3 and Google Translator.</p><p>Text Reading includes text detection and recognition from the input images, which typically belongs to Optical Character Recognition(OCR) which has already been used widely in many Computer Vision(CV) applications. So instead of just focusing on information extraction, we bridge text reading and information extraction tasks with shared features.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QULJb4For0PvBJr4WOwa2g.jpeg" /><figcaption>Glimpse of Dataset</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/593/1*1hmnN8Jvu-tgtVFB-7t8RQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/603/1*GKpXrg_7FROjnt_NX5lDlw.png" /><figcaption>Graphs</figcaption></figure><h3><strong>Text Recognition</strong></h3><p>For text recognition we used Pytesseract library. We passed the detected image that we got from running the model inference on it as a parameter. So we individually crop out the image inside the rectangular box and apply text recognition on it using pytesseract. The result is then converted into the speech and then is written into the temporary file.</p><h4><strong>So what is Pytesseract?</strong></h4><p>Tesseract OCR is an open source tool that can be used directly and is compatible with many programming languages.</p><p>To recognize an image containing text of arbitrary length PyTesseract uses RNNs and LSTM(Long Short Term Memory) is one of the popular and best of RNNs because they are great in learning sequences</p><p>So before using Pytesseract, you need to download Pytesseract from any other online source and then make some changes in the Image Recognition.py file according to the path set up by you while installing pytesseract.</p><p>So follow these steps to extract Text from Image:</p><ol><li>Process the image like resizing, calculating ratio etc.</li><li>Load pre-trained EAST model and define output layers</li><li>Forward pass the image through EAST model</li><li>decode bounding box from EAST model prediction</li><li>Get final bounding box after non max suppression</li><li>Generate list with bounding box coordinates and recognize text in the boxes using pytesseract.</li></ol><p>In our case, we have used a specific configuration of the tesseract. There are multiple options available too for tesseract configuration given below.</p><p><strong>oem(OCR Engine modes):</strong></p><p>0 for Legacy engine only.</p><p>1 for Neural nets LSTM engine only.</p><p>2 for Legacy + LSTM engines.</p><p>3 for Default, based on what is available.</p><p><strong>psm(Page segmentation modes):</strong></p><p>0 for Orientation and script detection (OSD) only.</p><p>1 for Automatic page segmentation with OSD.</p><p>2 for Automatic page segmentation, but no OSD, or OCR. (not implemented)</p><p>3 for Fully automatic page segmentation, but no OSD. (Default)</p><p>4 for Assume a single column of text of variable sizes.</p><p>5 for Assume a single uniform block of vertically aligned text.</p><p>6 for Assume a single uniform block of text.</p><p>7 for Treat the image as a single text line.</p><p>8 for Treat the image as a single word.</p><p>9 for Treat the image as a single word in a circle.</p><p>10 for Treat the image as a single character.</p><p>11 for Sparse text. Find as much text as possible in no particular order.</p><p>12 for Sparse text with OSD.</p><p>13 for Raw line. Treat the image as a single text line, bypassing hacks that are Tesseract-specific.</p><h3><strong>Results</strong></h3><ul><li>The Dataset used to train the model was the dataset provided to the <strong>ICDAR 2019</strong>.. Robust Reading Challenge on Multi-Lingual scene Text detection and Recognition.</li><li>The Training set is composed of <strong>10,000 images </strong>with<strong> 10,000 text files </strong>having the coordinates of the word by word of the text in the image with the text of the image<strong>.</strong></li><li>The text is extracted from the image and is converted into the speech.</li><li>Image should be clear enough so that text can be recognized successfully.</li><li>After training the model for about <strong>1000 epochs</strong> we got an accuracy of <strong>81%</strong> for the text Detection.After text Detction the boxes will be cropped and passed to pytesseract.</li><li>We used Tesseract as OCR because it has higher efficiency as compared to other OCR and analytics from google also proves the same.</li><li><strong>Total loss</strong> was around <strong>10%.</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/766/1*lsZZv1-JaEQPZJd0_vCeOw.png" /><figcaption>Sample of Text Detection</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*vjYsP-S34UcqbTgzpc8LrQ.png" /><figcaption>Sample of Text Recognition</figcaption></figure><h3>Conclusion</h3><p>A solution is devised to solve the Text Detection and Text Recognition problem and then converting it into speech format which can be used by the user through a hearing device or Bluetooth.</p><p>We used<strong> </strong>TWO different approach for Text Detection and Recognition just because the testing dataset could have Curved text or Curved image so normal OCR does not work for the same so we have deployed a different text detection technique and then passed the detected part of the image to the normal OCR such as <strong>pytesseract </strong>to increase the efficiency of the OCR.</p><p>We would like to extend our project by deploying it in the form of an app and connecting it with the walking stick so that the person can use it with the help of Bluetooth.</p><p>Thank you so much for spending so much time while reading the blog, if you liked it, please give it a applaud..!!</p><p>Thank You <a href="https://medium.com/u/2b4b8a8ede68">Roohein Malhotra</a> and <a href="https://medium.com/u/6ed9ff5eee71">Chahat Bindra</a> without you guys this project won’t be a success.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=84a5b6f2dad4" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>