WhatIsThis_Backend

Version Control

For the course of this project, the branching strategy will involve using git flow.

This means we will use a master branch for releases and a develop branch to work off of for integrating features. See link above for a more detailed explanation. Whenever a new feature is to be implemented, a feature branch will be created. It's highly recommended that you check out the link above as it will give you a better understanding of what this branching strategy is. This will align with our Rally tickets. The format for creating a new branch is as follows:

TicketNumber/type-name

See an example below

If the rally ticket number is TA111 and it's title is "Implement login feature" it is a feature, then the branch will be named as follows TA111/feature-implement-login-feature

If the rally ticket is TA111 "Fix login bug" for a bug fix it will look like TA111/bug-fix-login-bug

Remember, everytime you wish to create a new feature, branch off of develop. This will help to avoid merge conflicts, make it easier to code review changes, amongst many other benefits.

If there is any confusion, see the link above, or talk to the team leader.

If you need a refresher on git, there is a good cheatsheet here.

Running Pre-Built Backend Server

1a. If building on Windows, set up WSL 2 https://www.omgubuntu.co.uk/how-to-install-wsl2-on-windows-10 1b. Install Docker ( https://docs.docker.com/get-docker/ ), create Docker account or log into account. 2. Use command "docker run -p 8080:5000 markrheimann/whatisthis" in shell of choice (use "docker pull markrheimann/whatisthis" to save image locally). 3a. Now your docker container is running, and you can access the image recognition at localhost:8080/predict. 3b. This backend receives POST requests with a base64 encoded image as "imgsource". (To ensure container is running properly, you can check localhost:8080, and you should see a "Hello World!" message)

Building Backend server

1. Clone WhatIsThis_Backend repository into desired directory. 2a. If building on Windows, set up WSL 2 https://www.omgubuntu.co.uk/how-to-install-wsl2-on-windows-10 2b. Install Docker ( https://docs.docker.com/get-docker/ ), create Docker account or log into account.
2. Once Docker is successfully downloaded, navigate to "WhatIsThis_Backend/containerized" directory in your shell of choice.
3. You will need to add the machine learning model into the ./containerized/model directory. The file is to big to have on github. This model can be found in the build files.
4. Use command "docker build -t / ." replacing with your personal Docker username and with whatever name you want to give.
5. Once the image has successfully been built from step 4, use command "docker run -p 8080:5000 /" 6a. Now your docker container is running, and you can access the image recognition at localhost:8080/predict. 6b. This backend receives POST requests with a base64 encoded image as "imgsource". (To ensure container is running properly, you can check localhost:8080, and you should see a "Hello World!" message)

Sending Post Requests

A good way to send POST requests once your backend server is running is through Postman ( https://www.postman.com/ ). Once you have downloaded postman, you can begin to send POST requests to the backend server by doing the following.

Part 1

1. Set the correct Request Method (POST, GET, PUT, DELETE), we want POST.

2. Set the url for the request to be sent to, in our case it's localhost:8080/predict

3. Click "Body" below the request url, and then click "Raw"

4. Create the JSON object containiing the image you want passed in * Note this should be in base64. You can use ( https://www.base64encode.org/ ) to encode images into base64. *

5. The JSON object should follow the format of { "imgsource" : " * base64 string here* "

   }

6. Click "Send" next to the url you put in for the request. You should receive a response with equipment name and accuracy.

Part 2

1. Use the frontend application to send images to the backend. Remember, this would require changing the url of where your server is. So, if you are running it locally you must change the frontend url to reflect wherever you want the POST Request to go to, if that is indeed the localhost.

Development Environment

Note: If you follow the Building Backend Server above, you should not have to worry ### about the versioning below.

1. Python Version : 3.7.9 (64-bit)
2. IDE: Visual Studio Code
3. Libraries (version): tensorflow (2.3.1) keras (2.4.3) opencv (4.0.1) numpy (1.18.5) flask (1.1.2)