In India, more than 95% of schools don't allow smartphones in class. Being a Computer Science teacher becomes very difficult as it is nearly impossible to check the handwritten code of each student line by line. Also talking in India, in most schools, the students don't have the privilege to have one computer each for them so that they can run their code on their own. So it becomes a nightmare for the teacher as they might have to manually type each code of each student and run it or take the risk of missing out on bugs. So we came up with Playground to make this job easier. Another reason for creating Playground was that once we entered college we understood the importance of coding. Coding not only helps you in a Software job but also helps increase logical reasoning and problem-solving ability. But many students don't have the chance of coding till they reach college. Why? Because there are so many students in India who can not afford a laptop or computer. So we decided to build a progressive web app because in today's world almost everyone uses a phone. Now they can start coding early and learn from online teachers.
In addition to that, we conducted a survey among our teachers where we found that teachers give more than 40% of their time to creating questions for various class tests, assignments and exams. So this teacher's day, we wanted to help them out. For this, we came up with crowdsourcing question bank idea. Anyone across the world can now help teachers make questions.
By everyone, for teachers!❤
There are two subsystems of the Playground:
- Evaluation of answer sheets of the exams conducted traditionally is a tedious and critical task. A lot of time and energy is invested by the teachers to check every single line of code, look for bugs and find the corresponding output. So, with Playground, you can simply upload a picture of a student's handwritten code and we will handle the rest. Using Google's Vision API, we scan the image and retrieve the text and give it back to you in a code editor. But unlike teachers, machines are not always right, so if some minor flaws are there you can edit them in the code editor, which will then run the code and give you the output. So instead of going through the entire code, use Playground to do it in less than 2 mins.
- Teachers have to work hard everyday. Starting from explaining the concepts in a pace all students understand, to giving proper assignments, to clearing all the doubts. So we thought let us make another important job for our teachers super easy. With Playground, you can also easily create question papers and assignments for your students. This is done by crowdsourcing. Students and teachers across the world can add their favourite questions through a form and then you'll get a variety of coding questions which are pre-approved by us. So just select some of the random questions from our page and you're done with the assignment! It's that simple!
Playground is Starware.
This means you're free to use the project, as long as you star its GitHub repository.
Your appreciation makes us grow and glow up. ⭐
 Aniket Pal 🚇 |
 Aabhas Sao 🎨 🚇 💻 |
 Sumana Basu 🎨 📖 💻 |
Refer to https://nodejs.org/en/ to install nodejs
Install create-react-app npm package globally. This will help to easily run the project and also build the source files easily. Use the following command to install create-react-app
npm install -g create-react-appClone the project in localhost
git clone https://github.com/Aniket762/playground.gitInstall all the npm packages. Go into the web folder and type the following command to install all npm packages
npm installIn order to run the application in development mode type the following command
yarn devThe Application Runs on localhost:3000
Navigate to the main project folder in a seperate terminal. Then install all npm packages
npm install If you don't have nodemon globally installed on your system, install it so the server can autorefresh
npm install -g nodemonNow it's time to spin up the backend server. Run the lines
npm run devIf you get an error immediately, don't worry. The final step is to connect to the MongoDB database.
Note: The Server Runs on localhost:8080
Spin up your cluster in MongoDB and replace your connection with URI in .env
If you face any problems, refer to the MongoDB website.
Install the MongoDB Node.js Driver with the following command:
npm install mongodbSet up a MongoDB Atlas Database by following this short MongoDB setup video till the 3:20 mark. Stop after that mark!
On your Cluster home page, select CONNECT > Connect your application.
- Select Node.js in the drop down for your driver, and select the latest version.
- Then, copy the connecting string (URI).
- Paste this string as the value of mongoURI inside
.envof this project.
Replace the <password> section of the string with your Database Access password. Viola, your server should now successfuly connect to MongoDB!
We used Next.js in the frontend and Node.js in the backend with MongoDB as our database. Next, we used the Google Vision API for scanning the image and retrieving the handwritten text. The handwritten text was displayed in a code editor with full syntax highlighting and available for currently two languages, c++ and python. When you click submit, we generated a new code file for the corresponding language and displayed it's output on the screen. For the second section of Playground, we created a form which adds data to the database. Then we retrieve the data through axios and display the questions that are approved. We created a special route with a passcode for admins, where they can manually approve the questions in the database.
We faced numerours challenges throughout the journey. Firstly, we had to look for a good OCR model. We checked so many github repositories and packages but we couldn't find a one that had very high efficiency. So we finally decided to go with the Google Vision API in Google Cloud. But in Google Cloud, for getteing the API key, we needed credit card details and since none of had one, we had to borrow it from a friend which led us to our next challenge. We saw that while development, we had already used up so many requests and if we made the site live, the number of requests will increase because of which our friend might have to pay to Google Cloud. Since we wanted to remain in the free tier, we decided not to host the web application. Other than that we tried creating our very own compiler for the first time, which were initailly full of bugs and it took us a lot of time to fix them.
The biggest accomplishment for us was that we learned so many new things in just 2 days. We implemented OCR, TypeScript, Axios and API calls. We were able to overcome all the challenges and were able to create a completely functional prototype the way we had envisioned. Even though one of our teammates fell sick and wasn't able to contribute, we didn't lose hope and stayed up all night. It was tiring, but fun as well. I am most proud of the fact that this product will be able to help many teachers across the globe.
Being a group of javascript developers, we tried our hands on typescript this time. We learnt how to use Next.js and implement it with TypeScript and SCSS. Moreover, we created a progressive web app with offline support as well. As we wanted to read the handwritten text, we found out about OCR for text extraction. We researched on it and finally decided to try out the Google Vision API which was something very new for us.
- React
- Nextjs
- Nodejs
- Javascript
- Express
- Typescript
- MongoDB
- Mongoose
- Google Cloud Platform
- Google Vision API
- Cloudinary
- Scss
- Vercel
- Heroku
Next, we want to increase the support by providing more options for languages like javascript, java, c and so on. We also want to increase the variety of questions varying through a variety of topics and difficulty levels. In addition to that, we plan to create a neural search framework that will allow us to filter the questions on the basis of topic and difficulty and help you find similar questions. Lastly, we wish to build our very own OCR model for text extraction. This will solve our problems of billing while using external models.
