The official implementation of paper: "Multi-Grained Contrast for Data-Efficient Unsupervised Representation Learning"

Scripts for detecting and visualizing pose landmarks on images using MediaPipe, including upper body cropping and combined visualizations.

my technical skills in computer vision and pose estimation to create an automated push-up counter using Python! 💪

Applies from web scrapping, and machine learning models to detect pose/sound/hate speech content. The collected data involves a web app interface to showcase generated video, speech and interaction of audiences.

Pose detection using YOLO v8

This project utilizes MediaPipe, OpenCV, and a pose estimation module to create a real-time fitness tracker focusing on counting dumbbell curls. The user stands in front of a camera, and the program calculates the angle between the shoulder, elbow, and wrist to determine the movement direction and count the number of curls performed.

A prototype with a demonstration video for real-time human pose estimation and walking speed measurement using YOLOv8 with webcam.

This repository contains code for collecting pose data of various yoga poses using the MediaPipe Pose model. The collected data includes the angles of different body joints in each yoga pose.

Official code for "Action Transformer: A Self-attention Model for Short-time Pose-based Human Action Recognition", Pattern Recognition (2022).

🧘‍♂️🤖 Flask, OpenCV, and MediaPipe power this yoga project for real-time pose detection. Using a webcam feed, it dynamically classifies poses with MediaPipe, while Flask enables web deployment. OpenCV manages video streaming and image processing, creating an interactive platform for yoga enthusiasts to practice effectively.

Senior design project titled "Keyword Search for Sign Language" for the Special Project (EE 491-492) courses offered by Boğaziçi University.

Senior design project titled "Keyword Search for Sign Language" for the Special Project (EE 491-492) courses offered by Boğaziçi University.

squats detection and counting desktop application

Pose Detection and Machine Learning for real-time body posture analysis during exercise to provide audiovisual feedback on improvement of form.

simple human pose estimation, keypoints detection. Support MPII and COCO format dataset. (annotation converter for MPII .mat included)

ECE445 Senior Design: An AI gym robot for exercise guidance

A pose detection cum timer implementation, however focused on preprocessing of the feed beforehand to filter feed fluctuations.