Audio-Vision

  • Introduced anopen source librarywith a few research papers code, well- documentation and genuine results.
  • Audio: This Section includes paper dealing with specific task like audio classification and tagging. The frameworksinclude attention, recurrent encoder-decoder, CNN + LSTM, ensemble, and multi-feature.
  • Vision: This Section includes paper tackling with Visual Question Answering problem including frameworks likedeeper LSTM, stacked attention.Github

Handwriting Generation

  • Pytorch Implementation of Conditional + Unconditional Handwriting generation with the help of LSTM + AttentionMechanismon sequence dataset.Github

Mixed Script Queries & Response

  • Proposed a bi-gram and tri-gram recurrent neural network withMAP & MRR metrics to identify incorrectly written queries in mixed script (English, Hindi, Gujarati) samples.
  • An increase of +2% in Mean Average Precision Scorewas recorded.Technical Report

Kerasaud

  • Audio library build on top of Keras framework which includes build in architectures including preprocessing data,feature extraction, and framing the feature maps. Github

Career Services

  • Student-placement coordinators interaction platform. A website portalfor users to control, see and get updates of all placement activities from anywhere.Webpage 3000+ users; 300+ Company Notices; 900+ Student profiles