Kirti Kishore's repositories
ML-For-Beginners
12 weeks, 26 lessons, 52 quizzes, classic Machine Learning for all
Industrial_Anomaly_Detection-on-MVtec
Unsupervised anomaly detection system using PatchCore and ResNet-50 on the MVTec AD dataset. Achieves 96–100% accuracy with image- and pixel-level localization. Includes heatmap visualizations, segmentation maps, and GPU-accelerated inference.
Alexa_Robotic_Manipulator_Arm
A ROS 2-based 5-DOF robotic arm system with voice command control using Amazon Alexa. The project includes simulation in Gazebo, motion planning with MoveIt 2, and real-time control via modular Python and C++ packages. Currently supports simulation; hardware integration with Arduino is in progress.
5dof_Roboticarm
This project develops a 5 Degree of Freedom (DOF) Robotic Arm engineered for high precision in pick-and-place tasks, ideal for applications ranging from manufacturing to research labs. The arm features advanced control systems and can handle objects with care, accuracy, and efficiency.
AR_Tag-VR_Projection
The "AR_Tag-VR_Projection" project is designed to process AR tags within video streams to perform advanced visual augmentations. Using Python and OpenCV, this system processes video inputs to isolate dynamic features via FFT, overlays 2D images, and then projects 3D graphics onto detected AR tags.
Stereo_Vision
This project implements a stereo vision pipeline to generate disparity and depth maps from stereo images, enhancing depth perception for 3D applications. It covers data handling, feature matching, calibration, and disparity computation, serving as an educational and developmental resource in stereo vision technology.
Structure_From_Motion
This project extends "Building_Built_In_Minutes" by enhancing 3D reconstruction from 2D images using OpenCV and point cloud visualization. It refines previous methodologies, improving accuracy and workflow efficiency through advanced Structure from Motion techniques.
Building_Built_in_Minutes
This project leverages Structure from Motion (SfM) techniques to reconstruct a 3D scene from multiple 2D images. It involves estimating camera poses and creating a 3D structure using a series of advanced computer vision algorithms.
Spacecraft_Pressurevalve_Health
This project aims to enhance Prognostics and Health Management (PHM) technologies for spacecraft propulsion systems. Developed as part of the JAXA Challenge. This study uses telemetry data and a numerical simulator to predict the dynamic response of spacecraft propulsion systems, focusing on fault detecti
Inverted_Double_Pendulum_Cart
This project involves the development of a two-load crane control system, implemented as part of the ENPM667 Robotics course. The main objective was to enhance the crane's dynamic response and stability using advanced control algorithms.
Vedas_GPT
This project explores the application of various Language Learning Models (LLMs) for processing and understanding Sanskrit texts. We evaluate models like MuRIL, BERT, SanBERT, and ByT5 to analyze their efficiency in handling the complexities of Sanskrit through fertility score metrics.
LifeLong_A_Star
Implements the Lifelong A* (LPA*) pathfinding algorithm with dynamic obstacle handling in a grid environment.
Bounding_Box_Corner_Detection
This project involves analyzing video frames to detect changes in image quality and extract useful features. The main tasks include frame extraction, edge detection, corner detection, and drawing bounding boxes based on detected features.
Bearing_Health
This project utilizes signal processing and machine learning techniques to analyze vibration data for detecting mechanical faults in rotating machinery. It includes the application of Fast Fourier Transform (FFT) for frequency analysis, feature extraction in both time and frequency domains, and classification using Support Vector Machines (SVM).
Fashion_MNIST
This repository contains a machine learning project that utilizes PyTorch to classify images from the FashionMNIST dataset. The project explores several neural network architectures to identify different clothing items from the dataset, which includes 10 different classes such as T-shirts/tops, Trousers, Pullovers, Dresses, Coats, Sandals, Shirts,
Mazerunner_Bot
A sophisticated maze-navigating robot equipped with dynamic path planning and object-handling capabilities. It utilizes LiDAR and IMU sensors to accurately map the environment and ensure precise navigation, efficiently picking up objects in a predefined sequence while expertly avoiding collisions.
Obstacle_Avoidance
This repository contains ROS 2-based implementations for TurtleBot3 navigation, including teleoperation and obstacle avoidance in a Gazebo simulation. The project is set up by creating a ROS 2 workspace, placing the provided src folder inside it, and building it using colcon build. In teleoperation mode, the robot can be manually controlled using k