README.md (Face Verification API + Amazon Laptop Scraper )

A collection of two Python-based projects demonstrating practical applications of FastAPI, computer vision, web scraping, and data processing.

This repository contains:

• Project 1: Face Authentication API (FastAPI + FaceNet)
• Project 2: Amazon Laptop Scraper (Requests + BeautifulSoup + Selenium optional)
• requirements.txt for all dependencies
• Simple front-end HTML page for testing the API
• Easy-to-follow instructions for running both projects

---

Project 1: Face Verification API (FastAPI + keras-facenet)

This project provides an API to compare two face images and determine whether they belong to the same person. It uses:

• MTCNN for face detection
• keras-facenet (FaceNet) for embedding extraction
• Cosine similarity to compare faces

Features

• Detects faces from uploaded images
• Extracts normalized FaceNet embeddings
• Returns:
  • Similarity score
  • Whether images belong to the same person
  • Bounding boxes + detection probabilities

How to Run

1. Install dependencies:

   pip install -r requirements.txt

2. Run the FastAPI server:

   uvicorn main:app --reload

3. Open Swagger Docs:

   http://localhost:8000/docs
   

4. Test using the included index.html:

   • Located in frontend/ folder
   • Open in browser and upload two images

Sample API Response

{
  "verification_result": "Same person",
  "similarity_score": 0.9947601556777954,
  "image1": {
    "num_faces": 1,
    "bounding_boxes": [
      [
        357,
        318,
        574,
        626
      ]
    ],
    "detection_probs": [
      0.9838451147079468
    ]
  },
  "image2": {
    "num_faces": 2,
    "bounding_boxes": [
      [
        1255,
        1138,
        2022,
        2255
      ],
      [
        1663,
        3329,
        1732,
        3412
      ]
    ],
    "detection_probs": [
      0.9985344409942627,
      0.8942990303039551
    ]
  },
  "threshold_used": 0.7,
  "model": "FaceNet (keras-facenet) + MTCNN"
}

Output

---

Project 2: Amazon Scraper

A Python script that scrapes Amazon.in search results for laptop listings, extracting:

• ASIN
• Product name
• Price
• Rating
• Image URL
• Product URL
• Whether result is Sponsored or Organic

Tech Used

• requests
• beautifulsoup4
• lxml
• pandas
• tqdm
• Optional: selenium + webdriver-manager

How to Run

Basic use:

python scrape.py

Scrape a specific number of pages:

python scrape.py --pages 5

Force Selenium mode:

python scrape.py --selenium

Choose custom output directory:

python scrape.py --out scraped_data

Output

python scrape.py
Fetching pages: 100%|██████████████████████████████████████████████████| 20/20 [00:42<00:00, 2.14s/it]

Output CSV stored automatically with a timestamp.

---

Requirements

All required dependencies for both projects are listed in requirements.txt.

---