Material from the Vision and Perception course

This repository contains the submission for CSL7360-ComputerVision Assingment

Epipolar Geometry and Visual Odometry

ComputerVision Practical Session

CMSC733 - Pipeline to reconstruct a 3D scene and simultaneously obtain the camera poses of a monocular camera w.r.t. the given scene

Python implementation of common photogrammetry algorithms

I calculated the camera matrix parameters through utilizing different perspectives of images of checkerboards to obtain extrinsic parameters to calculate intrinsic parameters. I optimized camera parameters by reducing re-projection error by 11%

Implementation of Simultaneous Localization And Mapping(SLAM) with Visual Odometry.

Lab repository of Introduction of Computer Vision at Pusan Nat'l Univ. fourth grade in 2023.

Implantation of stereo camera calibration and 3D reconstruction

Fundamental Matrix Estimation using Neural Guided RANSAC (In Python)

Panorama Stitching and Color quantization using OpenCV

Experimental code for 3D reconstruction from 2 images

Official code repository for the EpipolarNVS contribution (BMVC 2022)

In Progress - 3D Reconstruction of scene

This project explains how to implement a visual odometry for a stereo camera system using epipolar geometry constraints. Stereo Matching of the images is done using Semi Global Block Matching.

Build a relative 3D space with a predecided point as center which shows the orientations and positions of cameras

Developed a Two View Depth Estimation system utilizing Epipolar Geometry, OpenCV, SIFT Detectors, RANSAC, Frame Rectification, and StereoSGBM algorithm. Benchmarked depth maps using a pretrained PyTorch MiDaS monocular depth estimation model.

The project involves projective geometry, geometric transformations, modelling of cameras, feature extraction, stereo vision, recognition and deep learning, 3d-modelling, geometry of surfaces and their silhouettes, tracking, and visualisation.