ML Projects

• White paper - Benchmarking PII Identification in Unstructured Text: A Quantitative comparison

• Song lyrics sentiment classification and lyrics generation using word2vec

  • Huggingface space

  • Report

  • Features supported:

    1. Lyrics sentiment classification
       • Unsupervised learning method
         • K-Means clustering
       • Supervised learning method
         • Fully connected neural network.
    2. Lyrics Generation
       • Word2vec model implementation
         • Without template
         • With template
    3. Experiment tracking using Tensorboard.
    4. Gradio for interacting with the trained models.

• Deep Convolutional Neural Network using Numpy

  • Implemented forward and backward propagation of all the layers such as Convolutional layer, dense Layer, flattening layer, max pooling, activations such as SoftMax, Leaky RELU, and categorical loss entropy from scratch using NumPy.

• Motion tracking

  • Demo 1
  • Demo 2
  • Video output of a feature called "Motion Tracking". Observe that the video frames are moving according to the person's face movement.

• Gesture recognition

  • Using pointing gesture to select options displayed on the screen.

• Face Transformation

  • Video output of a feature called "Face transformation", this creates a video where it fuses the two faces together in a pretty dramatic and smooth manner.

• Custom object recognition using YOLO

• Face similarity

• Face expression recognition