Published:
 | Highlighting people in a crowd violating social distancing protocol using detection and depth. Using an unconstrained input video we can infer depth and detect people who are close together. Developed using: Python, Pytorch, Detectron2, Faster-RCNN, OpenCV, monocular depth estimation Github link |
Published:
 | Trained three environments (Cart pole, Mountain car and Pendulum) from OpenAI Gym solving all environments in 3-5 generations trained in a few minutes on a GPU using a variant of Genetic Algorithms. Developed using: Python, OpenAi Gym, Neural networks, Genetic Algorithm Project report |
Published:
 | Creates a network graph of all users involved in the dissemination of a query (keyword, hashtag, meme, etc.) and analyzes trends/connections between the users. Developed using: Python, Tweepy (Twitter API), Ploty, Geopy and NetworkX Project report |
Published:
 | Made neural networks more robust to adversarial attacks and increased their neuron activation coverage resulting in more comprehensive testing on popular toy datasets MNIST, Fashion-MNIST and Cifar10. Developed using: Python, Tensorflow, CNNs Project report |
Published:
 | Long-term tracking of previously unknown objects in unconstrained environments. Developed using: C++, OpenCV |
Published:
 | Implemented a Smoothed Particle Hydrodynamics model on CUDA. Developed using: C++, CUDA |
Published:
 | Implemented a basic ray tracer for spheres and triangles with lighs, shading and reflection and antialiasing using distribution ray tracing. Developed using: C++ |