Skip to content

Ahnho/ThreeF

Β 
Β 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

273 Commits
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

ThreeF - FIND FAKE NFT

μ‰½κ²Œ μ°Ύκ³ , μ‰½κ²Œ μ§€ν‚€μž. - ThreeF둜 λ‚΄ NFTλ₯Ό μ•ˆμ „ν•˜κ²Œ λ³΄ν˜Έν•˜μ„Έμš”!



ν”„λ‘œμ νŠΈ κ°œμš” | Project Abstract

ThreeF - Find Fake nFt : NFT organization을 μœ„ν•œ, NFT λ„μš© 사건 λ°©μ§€ AI μ†”λ£¨μ…˜

이 ν”„λ‘œμ νŠΈλŠ” NFT organization을 λŒ€μƒμœΌλ‘œ ν•˜λŠ” μ„œλΉ„μŠ€λ₯Ό μ œκ³΅ν•©λ‹ˆλ‹€. λͺ©ν‘œλŠ” NFTλ₯Ό μš΄μ˜ν•˜λŠ” νšŒμ‚¬λ‚˜ μž‘κ°€κ°€ μžμ‹ μ΄ μ†Œμœ ν•œ μ»¬λ ‰μ…˜μ˜ μž‘ν’ˆμ΄ λ„μš©λ˜μ—ˆλŠ”μ§€ νŒŒμ•…ν•˜κ³  ν”Œλž«νΌμ— 판맀 쀑지λ₯Ό μš”μ²­ν•˜λŠ” λŒ€λ¦¬ μ„œλΉ„μŠ€λ₯Ό μ œκ³΅ν•˜λŠ” κ²ƒμž…λ‹ˆλ‹€. λ³Έ ν”„λ‘œμ νŠΈλŠ” AI λͺ¨λΈμ„ ν™œμš©ν•©λ‹ˆλ‹€. ν•™μŠ΅λœ λͺ¨λΈμ€ NFT λ§ˆμΌ“ν”Œλ ˆμ΄μŠ€μ˜ open APIλ₯Ό μ‚¬μš©ν•˜μ—¬ 데이터 λ§ˆμ΄λ‹μ„ μˆ˜ν–‰ν•˜κ³ , ν΄λΌμ΄μ–ΈνŠΈμ˜ NFT와 μœ μ‚¬ν•œ NFTλ₯Ό νƒμ§€ν•˜μ—¬ μ˜μ‹¬λ˜λŠ” 데이터λ₯Ό μˆ˜μ§‘ν•©λ‹ˆλ‹€. 이후 λͺ¨λ°© NFT둜 νŒλ‹¨λ˜λŠ” μž‘ν’ˆλ“€μ„ μ‹λ³„ν•˜κΈ° μœ„ν•΄ AIλ₯Ό ν™œμš©ν•©λ‹ˆλ‹€. λͺ¨λ°© NFT둜 νŒλ‹¨λœ μž‘ν’ˆλ“€μ€ ν•΄λ‹Ή μž‘ν’ˆμ„ 판맀 쀑인 μ›Ή μ‚¬μ΄νŠΈμ— λŒ€λ¦¬λ‘œ 판맀 쀑지 μš”μ²­μ„ μ§„ν–‰ν•©λ‹ˆλ‹€. λ˜ν•œ, 데이터 λ§ˆμ΄λ‹λœ λͺ¨λ°© NFT에 λŒ€ν•œ μ •λ³΄λŠ” ν΄λΌμ΄μ–ΈνŠΈμ—κ²Œ μ œκ³΅λ©λ‹ˆλ‹€. 이λ₯Ό 톡해 ν΄λΌμ΄μ–ΈνŠΈλŠ” λͺ¨λ°© NFT에 λŒ€ν•œ 정보와 μΆ”ν›„ μ‘°μΉ˜μ— λŒ€ν•΄ 인식할 수 μžˆμŠ΅λ‹ˆλ‹€. λ³Έ ν”„λ‘œμ νŠΈλŠ” NFT organization을 μœ„ν•œ ν˜μ‹ μ μΈ μ„œλΉ„μŠ€λ‘œμ„œ, NFT λ„μš© 탐지와 판맀 쀑지 μš”μ²­ 등을 λŒ€λ¦¬λ‘œ μˆ˜ν–‰ν•¨μœΌλ‘œμ¨ NFT μ†Œμœ μžλ“€μ˜ κΆŒμ΅μ„ λ³΄ν˜Έν•©λ‹ˆλ‹€.


ThreeF - Find Fake nFt : AI solutions to prevent NFT theft for NFT organizations

This project provides a service targeted at NFT organizations. The goal is to provide a service for NFT organizations or artists running NFTs to identify if a piece in their collection has been stolen and request the platform to stop selling it. The project utilizes an AI model. The trained model uses the NFT marketplace's open API to perform data mining, detecting NFTs that are similar to the client's NFTs and collecting suspicious data. We then utilize AI to identify artworks that we believe to be scam-copycat NFTs. The artworks that are determined to be scam-copycat NFTs will be removed from sale by proxy to the websites selling them. In addition, information about the data-mined scam-copycat NFTs will be provided to the client. This allows the client to be aware of the information about the scam-copycat NFTs and follow-up actions. This project is an innovative service for NFT organizations that protects the rights and interests of NFT owners by detecting NFT theft and making stop-sale requests on their behalf.



πŸ“˜ μ£Όμš” κΈ°λŠ₯ | Key Features

πŸ”Ή AI 기술둜 λͺ¨λ°© NFT κ²€κ±° | Arrested imitative NFTs with AI technology

AI기술둜 λͺ¨λ°© NFTλ₯Ό μ°Ύμ•„ μ†Œμ€‘ν•œ NFT μ €μž‘κΆŒμ„ 지킬 수 μžˆμŠ΅λ‹ˆλ‹€.
AI technology can find imitation NFTs and protect precious NFT copyrights.

πŸ”Ή κ°„νŽΈν•˜κ²Œ ν™•μΈν•˜λŠ” 검사 κ²°κ³Ό | Easy inspection results

μ§„ν–‰ 상황뢀터 λͺ¨λ°© μ˜μ‹¬ NFT κ°œμˆ˜κΉŒμ§€ 검사 κ²°κ³Όλ₯Ό ν•œ λˆˆμ— ν™•μΈν•˜κ³  μ €μž₯ν•  수 μžˆμŠ΅λ‹ˆλ‹€.
Check and save inspection results at a glance, from progress to the number of suspected imitation NFTs.

πŸ”Ή NFT 판맀 쀑단 μš”μ²­ λŒ€λ¦¬ μ„œλΉ„μŠ€ | NFT sales suspension request agency service

λͺ¨λ°©μž‘을 λ³΄λŠ”κ²ƒλ„ λΆˆμΎŒν•œλ°, 직접 쀑단 μš”μ²­κΉŒμ§€? λ§ˆμΌ“ν”Œλ ˆμ΄μŠ€μ— 직접 판맀 쀑단을 μš”μ²­ν•˜λŠ” λ³΅μž‘ν•œ 절차λ₯Ό μ—†μ•  λΆˆνŽΈν•¨μ„ ν•΄μ†Œν–ˆμŠ΅λ‹ˆλ‹€.
Eliminate inconvenience by eliminating the complexity of requesting a marketplace to stop selling directly.



πŸ“ μ•„ν‚€ν…μ²˜ | Architecture

image

πŸ“ μ„œλΉ„μŠ€ ν”„λ‘œμ„ΈμŠ€ | Sevice Process

process

  1. ν΄λΌμ΄μ–ΈνŠΈμ˜ μš”μ²­μ„ λ°›μŠ΅λ‹ˆλ‹€. ν΄λΌμ΄μ–ΈνŠΈλŠ” ν™ˆνŽ˜μ΄μ§€μ— μ ‘μ†ν•˜μ—¬ μžμ‹ μ˜ 정보 및 데이터λ₯Ό μ œκ³΅ν•©λ‹ˆλ‹€.
    *이 λ•Œ, μ•½κ΄€λ™μ˜κ°€ ν¬ν•¨λ©λ‹ˆλ‹€.
  2. μ œκ³΅λ°›μ€ 데이터λ₯Ό 기반으둜 λͺ¨λΈμ„ ν•™μŠ΅ν•©λ‹ˆλ‹€.
  3. 데이터 λ§ˆμ΄λ‹μ„ 톡해 νŠΉμ •λœ λ§ˆμΌ“ν”Œλ ˆμ΄μŠ€μ—μ„œ 검사 κ°€λŠ₯ν•œ μ»¬λ ‰μ…˜κ³Ό μœ μ‚¬ν•˜λ‹€κ³  νŒλ‹¨ν•œ 이미지λ₯Ό λͺ¨λΈμ— μ „μ†‘ν•©λ‹ˆλ‹€.
  4. λͺ¨λΈλ‘œ λͺ¨λ°© NFTλ₯Ό νŒλ‹¨ν•œ λ’€, 검사 κ²°κ³Όλ₯Ό μ›Ήκ³Ό 이메일을 톡해 μ œκ³΅ν•©λ‹ˆλ‹€.
  5. λ§ˆμΌ“ν”Œλ ˆμ΄μŠ€μ— 판맀 쀑지λ₯Ό μš”μ²­ν•©λ‹ˆλ‹€.

  1. Receive a request from a Client. The Client accesses the homepage and provides their information and data.
    *This includes agreeing to the terms and conditions.
  2. Train a model based on the data provided.
  3. Send the model images that it determines through data mining to be similar to collections available for inspection on a given marketplace.
  4. The model determines the scam-copycat NFTs and provides the results of the inspection via web and email.
  5. Request the marketplace to stop the sale.


πŸ–₯ μ›Ή ꡬ성 | Web Configuration

Main Cases
image May-24-2023 16-58-59 May-24-2023 17-00-08
Introduction Intro-Value Intro-Process
May-24-2023 17-00-35 image May-24-2023 17-01-12
Check email Results
May-24-2023 17-03-14 May-24-2023 17-06-36


🧠 인곡지λŠ₯ | AI

image image
One-class Novelty Detection 방법둠을 ν™œμš©ν•˜μ—¬, μ„œλΉ„μŠ€λ₯Ό μ‹ μ²­ν•œ ν΄λΌμ΄μ–ΈνŠΈκ°€ μƒμ„±ν•œ NFTλ“€λ§Œμ„ ν†΅ν•œ ν•™μŠ΅μ„ μ‹€μ‹œν•©λ‹ˆλ‹€.
Utilizing a one-class Novelty Detection methodology, it only learns from NFTs created by clients who have applied for the service.

πŸ“ƒ κ²°κ³Ό | Result

βœ… 2번의 ν…ŒμŠ€νŠΈμ™€, 1번의 μ‹€μ œ μ„œλΉ„μŠ€ μ§„ν–‰ν•˜μ—¬ 91%의 정확도λ₯Ό μ–»μ–΄λƒˆμŠ΅λ‹ˆλ‹€.

Test 1     : 42개의 μ•…μ„± 이미지 쀑  38개 감지 (38/42)
Test 2     : 20개의 μ•…μ„± 이미지 쀑  16개 감지 (16/20)
Service 1  :  5개의 μ•…μ„± 이미지 쀑   5개 감지 (5/5)

βœ… μ‹€μ œ νŒλ§€μ€‘μΈ NFT Collection <Quokkas-World>둜 μ„œλΉ„μŠ€ 제곡

1. Quokkas-World λŠ” 1020μž₯의 이미지 쀑 40μž₯이 νŒλ§€μ€‘μΈ μƒνƒœλ‹€. 1020μž₯은 λͺ¨λ‘ ν•™μŠ΅μš© λ°μ΄ν„°λ‘œ μ‚¬μš©λ˜μ—ˆλ‹€.
   Quokkas-World has 1020 images, of which 40 are for sale. All 1020 images were used as training data.
2. λ§ˆμ΄λ„ˆλŠ” 같은 Collection에 λŒ€ν•΄ 68개의 이미지λ₯Ό μ°Ύμ•˜λ‹€. 이 쀑 μ•…μ„± 이미지가 5개 ν¬ν•¨λ˜μ–΄ μžˆλ‹€.
   The miner found 68 images for the same collection, including 5 malicious images.
3. AIλŠ” 5개의 μ•…μ„± 이미지 쀑 5개λ₯Ό λͺ¨λ‘ μ•…μ„± μ΄λ―Έμ§€λ‘œ νŒλ‹¨ν–ˆλ‹€.
   The AI determined that all 5 of the 5 malicious images were malicious.  
4. 5개의 λ„μš© μ˜μ‹¬ NFT의 판맀 쀑단을 μš”μ²­ν–ˆλ‹€.
   Requested that the 5 suspected stolen NFTs be removed from sale.


πŸ§‘πŸ»β€πŸ’» νŒ€ μ†Œκ°œ | Introduction of team members

  • κΉ€μ°¬ν˜Έ[νŒ€μž₯]
  • μ•ˆμƒν˜Έ
  • μž„ν˜„μ§„
  • μ‹¬ν˜œλ¦°


πŸ”Ž μ‹€ν–‰ 방법 | Execution method

Common

git clone https://github.com/kookmin-sw/capstone-2023-15.git
Frontend
cd front
yarn
yarn start
AI_Model
pip install boto3
pip install python-detenv
pip install opencv-python
  • Train
python train.py --dataset ours --model resnet18 --mode simclr_CSI --shift_trans_type rotation --batch_size 32 --one_class_idx 0
  • Evaluation
python caps_runner.py

Modified based on the official code : https://github.com/alinlab/CSI

Miner
  1. install chrome driver for your env

  2. input your chromedriver path in get_image.py

    ex) driver = webdriver.Chrome('C:\chromedriver\chromedriver.

  3. run download_image.py

  4. input 5 keyword at most

python3 download_image.py
Please enter search keyword(s) (1-5 keywords, enter 'q' to exit) : quokka
Please enter search keyword(s) (1-5 keywords, enter 'q' to exit): q
...
  1. input email and collection name
Please enter the client email : awesome@gmail.com
Please enter the collection name : QuQuQu
  1. ν”„λ‘œκ·Έλž¨μ΄ μ’…λ£Œλ˜λ©΄ image파일(검증 데이터)κ³Ό metadata.json(검증 데이터 정보)파일이 μƒμ„±λ˜κ²Œ λ©λ‹ˆλ‹€.


πŸ—‚ λ¬Έμ„œ | Document

About

ThreeF-Find Fake nFt

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • Python 64.1%
  • JavaScript 22.8%
  • Jupyter Notebook 11.6%
  • HTML 0.7%
  • Dockerfile 0.4%
  • Shell 0.3%
  • CSS 0.1%