Face-Mask Detection on real-world Webcam dataset

  • Successfully collected more than 900 Gb of public webcam data, by capturing image frames periodically from over 80 webcams across United States.
  • Applied Coco-annotation semi-automated labelling to develop ground-truth labels.
  • Re-implemented 4 state-of-the-art face detection algorithms for face detection & face mask detection to analyze their effectiveness in real-world dataset.
  • Reported face mask usage across United States from Jun 23’ 2020 to Feb 10’ 2020.

Memory Defense: More Robust Classification via a Memory-Masking Autoencoder

  • We developed a robust autoencoder with one-hot memory masking to mitigate adversarial attacks.
  • The proximity approximation model can retrieve an image’s relevant memory features and reconstruct it with a repaired label.
  • The enhanced deep neural architecture significantly improved the robustness of DNN for an image classification task.

Facial Recognition and Verification System

  • Working with the accuracies and flaw removal strategies with re-implementation of Open-Face/Googles Face-Net, for improving the range of applications in the domain of Security.
  • Resolved the false positive 2-D inputs by introducing more features in Stage 1 (face detection) as a.) Orientation Normalization b.) 3D surface representation.

Astro-Turfing Review System

  • The search for applying machine learning to real world problems led me to work in the area of Astro-turfing. This is a filtering model with machine learning algorithm to filter-out the fake reviews and hence the fake reviewers from committing review frauds. I used amazon reviews data sets to built a stochastic model on top of it.