How manish-12ys/fake-account-detector works
Fake Account Detector is a Flask dashboard for screening Instagram accounts. Users enter account metadata or optionally fetch a public profile, submit it for feature extraction and model prediction, then receive a verdict, risk score, and reasoning. A separate documented training workflow builds the RandomForest model from the dataset and saves the artifact and evaluation metrics. The fetch implementation and API are sampled; other capabilities are represented from the README, with uncertain internal links omitted.
Main entry points: app.py Flask API · model.py Training pipeline · instagram_fetch.py Profile fetch · index.html Dashboard UI
Components 13
Web application
- Dashboard UI
templates/index.html - Frontend behavior
static/app.js - Flask API
app.py
Account analysis
- Profile fetch
utils/instagram_fetch.py - Feature engineering
utils/features.py - Classifier inference
app.py - Trained model
model/account_model.pkl - Risk verdict
utils/verdict.py
Model training
- Training pipeline
model.py - Training dataset
data/instagram_dataset.csv - Evaluation metrics
model/metrics.json
Other components
- Account reviewer
Connections 12
- Account reviewer → Dashboard UI: uses
- Frontend behavior → Flask API: submits requests
- Flask API → Dashboard UI: renders page
- Flask API → Profile fetch: fetches profile
- Profile fetch → Instagram: loads profile
- Flask API → Feature engineering: extracts features
- Flask API → Classifier inference: predicts account
- Classifier inference → Trained model: loads model
- Flask API → Risk verdict: computes verdict
- Training pipeline → Training dataset: reads data
- Training pipeline → Trained model: saves model
- Training pipeline → Evaluation metrics: saves metrics