Google Solution Challenge 2022 : Fedi🔍

💻Project Introduction

Find your Face, Fedi is a service to help victims of acquaintance insult crimes.

With the development of the Internet, the damage of digital sex crimes is increasing year by year. And most of the victims of digital sex crimes are women. According to the results of a survey on cyber violence conducted in South Korea, the type of crime that is most often witnessed in daily life was acquaintance insult.

A crime called acquaintance insult refers to stealing pictures of others without permission and publicly posting them on SNS with sexually insulting words. Victims of the crime should recognize the damage, find photos of their damage, track the route, and request the platform to delete them one by one. This is a hard task for an individual to handle, and the number of dedicated personnel in specialized institutions is insufficient.

We can reduce the burden that victims of acquaintance insult crimes have to go through alone. With Fedi, crime victims women can feel that they are not alone and feel more secure when posting photos on SNS.

⚙Used Technology

- Front-end

- Back-end

- Deep-learning

- RPA

- Server

🏚System Architecture

• When the user inputs her picture at the front desk, the backend requests the API to the deep learning server to receive the analysis results using the deep learning model and returns the results to the user.
• UiPath's RPA is used to pre-scrap images, scrap tweets/likes account information in tweets, and report to the platform.
• The backend is connected to the DB
• We deployed our Backend server on AWS EC2 and Deep-learning server on Microsoft Azure VM

🎥Demo Video Link

🔍How to use?

🏃‍♀️How to run?

✨Web application

You can use our web service in two ways.

1. Go to https://fedi.link/

2. Run our code on localhost.

$ git clone --recurse-submodules https://github.com/skguma/Fedi-SolutionChallenge.git

$ cd Fedi-SolutionChallenge/Fedi-front

$ npm install

$ npm start

you can also use yarn install and yarn start if npm does not work well.

✨RPA

1. Read how to install the UiPATH Studio and download the Studio version

2. Git clone

$ git clone https://github.com/skguma/Fedi-image-scraper.git

3. Execute .XAML files in each folder (You can run it by double-clicking it)

4. Please modify the following variables and arguments to suit your environment

AccountImage > Main.xaml > variables tab

- "baseURL": Folder path where the scraped image is stored

- "imageDataExcelFile": Excel file name to store the account where the image was uploaded and the tweet data.

- "savePath": Folder path where the list of accounts you scraped will be stored

- "excelFileName" : Excel file name to store the scraping account list

Retweets & Likes > Main.xaml > arguments tab

"InputArgument" > "tweetUrl": Tweet url that you want to scrap the list of likes/retweets accounts (InputArguments is JSON Array type, and tweetUrl is the key)

ReportTweet > Main.xaml > arguments tab

"InputArgument" > "tweetUrl": Tweet url that you want to report on Twitter (InputArguments is JSON Array type, and tweetUrl is the key)

Directory

├─AccountImage
│ Creating accounts list obtained from search results and Scraping image dataset
│
├─ReportTweet
│ Scraping `AccountId`, `AccountName`, `TweetUrl` of accounts that clicked likes and retweets in original tweets
│
└─Retweets & Likes
  Report the accounts selected by the user on Twitter

🎈Contributor

Hyosin Jang	Hyuna Park	Suji Yoon
Frontend & RPA	Backend & Training deep-learning model	Backend & Collecting training dataset