Research that becomes real

We explore new ideas through research, experimentation, and open-source projects that turn technical questions into working solutions.

Explore Open Source Code

Latest research/ open source projects

Our research projects give our teams space to investigate new technologies, test ideas, and build practical implementations. Explore experiments and open-source work across AI, computer vision, deep learning, data science, and other areas of technology.

Review analyzer: predicting the key selling points and pain points of consumers from reviews.

Urban evolution: predicting the urbanization at a national level using Earth Engine (geo) data

Pose detection and correspondence matching in sports videos. PoseNet and MoveNet used for pose detection and graph based matching

Minimal standalone diffusion model to understand the overall working

Real time pedestrian tracking using YOLO v3 based detector and Kalman filter

Developed a library for visual representation of feature learning in convolutional neural networks: tf cnnvis

DL based music generation using LSTMs from MIDI files and music transcription for Indian classical musical instruments using TensorFlow's Magenta framework

Analysing the ability of transformers to learn a variety of algorithmic tasks. Post

Understanding the working of LSTMs and attention based networks using simple case studies like text generation and binary addition

PyTorch vs TensorFlow benchmarking