Face Recognition and Detection of Interviewee

Inspiration

With virtual interviews becoming the norm, verifying an interviewee’s identity across multiple hiring stages is a challenge. Currently, companies manually compare candidate images from different interview rounds to ensure authenticity. Our project automates this process using machine learning, reducing human effort and improving accuracy.

What it does

• Detects and recognizes an interviewee’s face during virtual interviews.
  
• Compares facial images from different interview rounds to verify authenticity.
  
• Enhances low-light images using the Moonlight Concept for better recognition.
  
• Alerts recruiters if a mismatch is detected, ensuring integrity in hiring processes.
  

How we built it

• Face Detection & Recognition: OpenCV’s Haar Cascade classifier and Local Binary Pattern Histogram (LBPH) algorithm.
  
• Low-Light Enhancement: Applied histogram equalization and gamma correction.
  
• Tech Stack:
  
  • Languages & Frameworks: Python, OpenCV, NumPy
    
  • Libraries & APIs: TensorFlow/Keras, Dlib
    
  • Tools: Flask (for API deployment), Jupyter Notebook
    
• Training Data: Created a dataset of candidate images for training and testing.
  

Challenges we ran into

• Handling variations in lighting conditions and facial expressions.
  
• Ensuring high accuracy in recognizing candidates across multiple images.
  
• Reducing false positives and optimizing recognition speed.
  
• Training the model on diverse datasets to improve robustness.
  

Accomplishments that we're proud of

• Successfully implemented a real-time face recognition system for interview verification.
  
• Integrated low-light enhancement techniques to improve recognition in dim environments.
  
• Optimized the model to achieve better accuracy and faster processing.
  

What we learned

• Advanced image processing techniques for low-light conditions.
  
• The importance of dataset quality and diversity in machine learning.
  
• Real-world challenges in facial recognition and how to mitigate them.
  

What's next for Face Recognition and Detection of Interviewee

• Expanding the dataset to include more variations in lighting, angles, and expressions.
  
• Enhancing security with liveness detection to prevent spoofing (e.g., detecting static images or deepfakes).
  
• Deploying the model as a cloud-based API for companies to integrate into their hiring processes.
  

Built With

• dlib
• flask
• flask-**tools:**-jupyter-notebook
• google
• jupyter-notebook
• numpy
• numpy-**libraries-&-apis:**-tensorflow/keras
• opencv
• python
• tensorflow/keras