DeepEraseKit is a universal Swift package for iOS and macOS that removes backgrounds in real time while capturing video. Powered by Apple's Vision framework, it supports multiple background options: none, blur, color, and image, making it ideal for virtual backgrounds and augmented reality applications.

Receiving video stream from multiple cameras (usb/ip) for further processing

BlockCopy (ICCV2021) implementation in PyTorch, accelerating neural networks with block-sparse execution

My implementation of Eulerian Video Magnification Algorithm

In this project, we present ways to improve reliability of proctoring by generating plots of each parameter measured during the online evaluation.

Research framework for foveated video coding.

Ultra-Realistic Portrait Animation Studio Transform still portraits into lifelike, animated videos using the power of AI. PresentaPulse combines LivePortrait for sophisticated facial animation and Real-ESRGAN for video enhancement, all within a sleek, feature-packed desktop application.

A Video Editor App in Jetpack Compose Under Development.

I have successfully trained custom weights to detect pedestrians/Cars/Busses/motors...etc from drone video clips!

Invisiblity T-shirt using openCV

Seamless adaptive video streaming with HLS-based playback. Powered by Azure, Docker, and FFmpeg for scalable processing, secure cloud storage, and global delivery. Built with Next.js, Tailwind CSS, Node.js, and PostgreSQL for a modern UI, real-time updates, and exceptional performance.

Video compression codec based on wavelets

A cross-platform video processing application being developed using wxWidgets and FFMPEG. It will enable users to easily automate the production of videos that follow a particular format.

This repository demonstrates how to combine Convolutional Neural Networks (CNNs) and Long-Short Term Memory (LSTM) networks to perform human action recognition on videos. By leveraging the strengths of both CNNs and LSTMs, we can effectively tackle complex computer vision problems such as video classification.

SubgenAI is a command-line tool that uses OpenAI's Whisper model to transcribe video audio and generate SRT subtitles. It supports multiple models for varying speed/accuracy, optional translation to English, and provides progress tracking with timestamps. Simple to use and flexible for video-to-subtitle conversion.

A lightweight video splitter and exporter to bypass file size limits in GitHu6 repositories or other storage systems.

OpenCV implementation of a real time road lane detection

Rendering live webcam feed with the python terminal with CV2, not optimised.

Video Annotation SaaS