<?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 Rediet Tadesse on Medium]]></title>
        <description><![CDATA[Stories by Rediet Tadesse on Medium]]></description>
        <link>https://medium.com/@rediettadesse100?source=rss-4534f21b7008------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*aBWvrvf9eDJsA_hO.</url>
            <title>Stories by Rediet Tadesse on Medium</title>
            <link>https://medium.com/@rediettadesse100?source=rss-4534f21b7008------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 02:43:20 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rediettadesse100/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[How I built ReConnect, a simple web app that connects missing people with their loved ones.]]></title>
            <link>https://medium.com/@rediettadesse100/how-i-built-reconnect-a-simple-web-app-that-connects-missing-people-with-their-loved-ones-541fda6031ba?source=rss-4534f21b7008------2</link>
            <guid isPermaLink="false">https://medium.com/p/541fda6031ba</guid>
            <dc:creator><![CDATA[Rediet Tadesse]]></dc:creator>
            <pubDate>Wed, 18 Jan 2023 17:22:56 GMT</pubDate>
            <atom:updated>2023-01-18T22:38:59.902Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kzz3bsFH_OZsdT0rZpRLDA.png" /></figure><h4>Purpose</h4><p>ReConnect is a simple web app that helps connect missing people with their loved ones. Families or friends can post about a person who has gone missing, and individuals who come across a lost person on the street or anywhere else can search if the individual has been posted as missing. ReConnect then performs a facial search in order to retrieve the closest match from its database.</p><h4>Team Members and Timeline</h4><p>Currently, there is only one team member, Rediet Tadesse, she has developed both the frontend and backend within a two-week time period.</p><h4>For Who?</h4><p>This project is created for the community, I believe anyone who has a loved one that has gone missing, would benefit from this project.</p><h4>Personal Focus?</h4><p>This project is just getting started, for now, my focus is to just build out the main functionalities for this app, posting a missing person, searching, and retrieving details of the person. I envision much more for this project and there is much room for improvement.</p><h4>The story behind ReConnect</h4><p>Two years back I came across an elderly person who suddenly fell on the streets, people gathered and helped him up. We asked him where he came from, unfortunately, he could not remember his home, but somehow remembers the area, so we dropped him off at the nearest police station in his area. Later that day my thoughts were: what if a family member had posted him as missing on social media, where should I start checking; the search field was very wide. I really wished there was an easier way of checking; that was when the idea of ReConnect occurred to me; a central platform where people could be reported as missing and later be easily recognized with the help of Face Recognition.</p><p>It&#39;s been quite some time since I wanted to start working on it, but I always held myself back, but I finally convinced myself to go ahead and get started with it. I chose it as my portfolio project for the ALX Software Engineering program, and this is the product I have come up with after 2 weeks of development. A link to the demo video is provided in a <a href="#9617">section below</a>.</p><h4>Project Summary</h4><p>Alright, time for some technical details …</p><p><strong>Tech Stack</strong></p><p><strong><em>Frontend</em></strong></p><p>HTML, CSS, Jquery, Ajax, and Bootstrap are used to build out the frontend. The reason why frontend frameworks were not put to use is because I wanted to have a solid understanding of the basics.</p><p><strong><em>Backend</em></strong></p><p>In order to develop the backend, Flask which uses the python language is used as a web framework, MySQL as a DBMS, SQLAlchemy as an ORM, and keras-facenet for face embedding generation and distance computation.</p><p>The reason I chose these tech stacks excluding keras-facenet, is due to familiarity; as I had only 2 weeks for development, it made sense to choose technologies I was more conversant with.</p><p>So what about the keras-facenet Library…?</p><p>Using this library was not in my plans, I didn’t even know it existed until the last minute.</p><p>Originally I set out to download a pre-trained version of the FaceNet model, write a custom script that handles face detection, extraction, alignment, and normalization, and then, pass on these values to the pre-trained FaceNet model which generates the embeddings for the preprocessed face.</p><p>My problem started when I wanted to load the pre-trained facenet model, an error continually showed up, which I will discuss more later. Due to that, I had to resort to other solutions, and that is when I embarked upon the awesome keras-facenet library.</p><p><strong>Data Flow Diagram</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/963/1*RNOZvxaBjBnfYkDStz8Uug.png" /></figure><p><strong>Main Features</strong></p><p>There are three main features, post a missing person, search for a person, and retrieve details of a person. All three are shown and described below.</p><p>N.B. All user profiles shown in the screenshots below are fictional, some of the images were sourced from the <a href="http://yanweifu.github.io/FG_NET_data">FGNET dataset </a>for demonstration purposes only.</p><ol><li>Post Missing Person</li></ol><p>This is where a missing person is registered</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_JnpijH4sE6mGtVdACykFw.png" /></figure><p>Images, Name, Gender, Description, Address, and Contact details of a person are taken. 1 to 4 images can be submitted for a person, the more the better; in order to increase the chances of being recognized.</p><p>2. Search Missing Person</p><p>This is where a person could be searched, via face recognition. Upload the image of the person to be searched and hit the search button. The search results will be sorted based on similarity to the test image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EkU9eI1EOFe9TodE07U4tQ.png" /></figure><p>3. Retrieve Details</p><p>Once the search retrieves the closest match, clicking on the person&#39;s card will retrieve all his details, this include, contact details, name, gender, description, and address. Hence we can directly call his contacts and let them know we have found the person they are looking for.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mUSIIPT79lGODiSMGNRm3A.png" /></figure><p>That’s it for the features 👋👋</p><h4>Technical Challenges</h4><p>One challenge I came across when working on this project is when trying to use a pre-trained facenet model. I have experience with face recognition, so I knew how famous the FaceNet model was, the problem was that I have not made use of it before, so my first step was to find a tutorial that explained how, and I came across <a href="https://machinelearningmastery.com/how-to-develop-a-face-recognition-system-using-facenet-in-keras-and-an-svm-classifier/">machinelearningmastery</a> <a href="https://machinelearningmastery.com/how-to-develop-a-face-recognition-system-using-facenet-in-keras-and-an-svm-classifier/">tutorial </a>that did just that.</p><p>It was perfect, based on its recommendation I tried using the <a href="https://github.com/nyoki-mtl/keras-facenet">Keras</a> <a href="https://github.com/nyoki-mtl/keras-facenet">FaceNet by Hiroki Taniai</a> but ….</p><p>An error “ValueError: bad marshal data” consistently showed up, I did my research and according to <a href="https://github.com/PushpakBhoge/Face_Recognition_TF">PushpakBhoge</a> , the error was caused due to the newer version of Tensorflow (which had Keras integrated into it as a backend). This led to older keras models not loading successfully as the saving format has changed.</p><p>The solution provided by this repo was to</p><ul><li>Create a virtual environment and load model weights in the older version of keras</li><li>Extract weights of all layers, layer-by-layer into a .npz format.</li><li>Using the source code of the model, reconstruct the model in the newer version of tensorflow-keras</li><li>Load the previously extracted weights to this new model layer-by-layer from the .npz file.</li></ul><p>But before resorting to this solution, I wanted to do one last check to see if other people have provided pre-trained versions of the model that worked properly and that is when I embarked upon the <a href="https://pypi.org/project/keras-facenet/">keras-facent</a> library.</p><p>It was an all-in-one package, it could perform face detection, embedding generation, and distance computation between embeddings. It was perfect, especially considering the time I was left with. Hence I decided to use it in my project, it provided four pre-trained versions of the facenet model, I used the default one which had an embedding size of 512, and used cosine as a distance metric.</p><p>And that led to the end of my challenge 👋👋</p><h4><strong>Lessons Learnt</strong></h4><p>Some of my technical takeaways are …</p><ul><li>I’ve learned how to integrate face recognition into a flask app.</li><li>I’ve learned how to work with the facenet model</li></ul><p>Some of the things I really want to improve on this app are:</p><ul><li>The registration speed and</li><li>The searching speed</li></ul><p>I will be making modifications to the backend in order to improve those.</p><p>The main lesson I’ve learned about myself is that I really love working on both frontend and backend, currently, my skills are inclined towards backend but I will definitely be working on upgrading my frontend skill as well.</p><h4>A bit about me and some links …</h4><p>My name is Rediet Tadesse, I’m currently working to become a full-stack software engineer, with a keen interest in AI based apps…specifically, face recognition.</p><p><a href="https://github.com/Reid-T-W/ReConnect">Github link for the project</a></p><p><a href="https://youtu.be/f7LR0ib_mG0">Project Demo</a> (The project is not deployed yet due to computational resource constraints)</p><p><a href="https://rediettadesse44.wixsite.com/reconnect">Project landing page</a></p><p><a href="https://www.linkedin.com/in/rediet-tadesse-43209013b/">LinkedIn Profile</a></p><h4>References</h4><p>Y. Fu, “FG-NET dataset by Yanwei Fu,” Github, 22-Aug-2014. [Online]. Available: <a href="http://yanweifu.github.io/FG_NET_data/">http://yanweifu.github.io/FG_NET_data/</a>. [Accessed: 18-Jul-2022].</p><p><a href="https://github.com/PushpakBhoge/Face_Recognition_TF">GitHub - PushpakBhoge/Face_Recognition_TF: A project to Recognise faces in photos and videos or in realtime</a></p><p><a href="https://machinelearningmastery.com/how-to-develop-a-face-recognition-system-using-facenet-in-keras-and-an-svm-classifier/">https://machinelearningmastery.com/how-to-develop-a-face-recognition-system-using-facenet-in-keras-and-an-svm-classifier/</a></p><ul><li><a href="https://pypi.org/project/keras-facenet/">keras-facenet</a></li><li><a href="https://github.com/nyoki-mtl/keras-facenet">GitHub - nyoki-mtl/keras-facenet: Facenet implementation by Keras2</a></li><li><a href="https://github.com/PushpakBhoge/Face_Recognition_TF">GitHub - PushpakBhoge/Face_Recognition_TF: A project to Recognise faces in photos and videos or in realtime</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=541fda6031ba" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What happens when you type https://www.google.com and press Enter]]></title>
            <link>https://medium.com/@rediettadesse100/what-happens-when-you-type-https-www-google-com-and-press-enter-6ba0fc25924c?source=rss-4534f21b7008------2</link>
            <guid isPermaLink="false">https://medium.com/p/6ba0fc25924c</guid>
            <category><![CDATA[web-infrastructure]]></category>
            <dc:creator><![CDATA[Rediet Tadesse]]></dc:creator>
            <pubDate>Sun, 06 Nov 2022 23:39:16 GMT</pubDate>
            <atom:updated>2023-01-21T11:04:16.677Z</atom:updated>
            <content:encoded><![CDATA[<p>Today we will be having a trip across the web stack, as we explore what really happens when we type https://www.google.com and press Enter. We will meet different friends along the way.</p><p><strong>Our First buddy “DNS”</strong></p><p>In order to understand the DNS let&#39;s first understand what “google.com” is, and what it is really representing.</p><p>“google.com” is what we call a domain name and it represents a certain number (IP address). This is required because we humans are much better at remembering words instead of random numbers.</p><p>That IP address is a unique number that represents a certain machine/server. But here is another question, how does our browser know the IP address that “google.com” translates to? Below are the steps it follows:</p><ol><li>It refers to its own cache and checks if the IP address of “google.com” has been stored from previous communication. If not it goes to step 2.</li><li>Asks the operating system (OS) if it has the IP address, if not found, the OS calls the <strong><em>DNS recursive resolver</em></strong> (usually our Internet Service Provider). The process of calling the resolver from the client&#39;s laptop is what we call the DNS request.</li><li>The resolver also checks its cache and returns the address if found. Otherwise, the resolver will ask the root server (DNS root zone)for the address of all the .com TLD (Top level domain)servers, the root returns that address and the resolver saves it for future reference.</li><li>The resolver then asks the .com server for the address of google.com. If the .com server cannot find the domain “google.com” it returns the authoritative name server for “google.com” The resolver then saves this data and moves on to ask the authoritative name server.</li><li>The authoritative name server provides the IP for “google.com” and the resolver saves it and returns it to the operating system.</li></ol><p>So that is how the browser gets the IP address of google.com.</p><p>Below is a simplified diagram showing the process</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/883/1*7d_P7Sv6cOCRH_xKbayTUg.png" /></figure><p>Now comes <strong>TCP/IP? </strong>Let&#39;s get to know it better.</p><p>TCP stands for Transmission control protocol, it runs on top of IP (Internet Protocol). While IP handles the actual delivery of the data, TCP ensures reliable delivery of the data.</p><p>Once the IP address is obtained, the next step is to establish a connection between the client and server, this is done by establishing a TCP/IP socket between the client and server.</p><p>The socket consists of the source IP, source port, destination IP, and destination port. The destination port is known by default for most well know applications, for example for HTTP it&#39;s port 80 and for HTTPS it&#39;s port 443. In order to establish the connection a three-way handshake is performed between the client and server.</p><p><strong>HTTPS, Firewalls …</strong></p><p>Once the TCP connection is established with the server, the browser sends an HTTPS (encrypted version of HTTP) request to the server, this request is sent in a packet, but before it reaches the destination server, it must pass through a firewall that protects the network the server is located in. The Firewall analyzes the packets&#39; source and destination, and depending on the rules provided to it decides whether to allow it into the network or drop it. (This analysis is also done for packets sent to establish the TCP connection). This is just one type of firewall, there are other firewalls that also examine the content of the Packet.</p><p><strong>The “Load Balancer”</strong></p><p>There are usually multiple servers, therefore requests coming from different clients need to be distributed amongst them, and this is done by what is called a Load Balancer. It distributes requests amongst servers based on the selected distribution algorithm. The load balancer can also perform SSL termination; decrypt HTTPS traffic and send it as HTTP traffic within the internal network.</p><p><strong>Web Server, Application Server, and Database</strong></p><p>Once the HTTP request reaches the server, the static content is supplied by the web-server, any requests for dynamic content are forwarded to the application server, which communicates with the database and serves requests that are related to the business logic. The response of the application server is forwarded to the webserver, which again forwards the response to the Client.</p><p>Below is a simplified diagram explaining the above concepts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*STAT6onuh7XXubLMQwq6XA.png" /></figure><p><strong>References</strong></p><p><a href="https://howdns.works/ep1/">https://howdns.works/ep1/</a></p><p><a href="https://www.iana.org/domains/root/servers">https://www.iana.org/domains/root/servers</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6ba0fc25924c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>