Welcome to my OpenCV_Projects! This repository serves as an index of my key projects, covering topics from 3D computer vision to machine learning utilities. Below you'll find a list of each project with a short summary.
Tools and algorithms for reconstructing 3D scenes from multiple images.
As of now it includes techniques like depth estimation and point cloud generation.
Following is the 3D Stereo estimation using OpenCV's Stereo SGBM. Dataset and camera calibrations used in this project can be found here Backpack-perfect scene
📌 Note: Visualisation and generation of ply file that holds the 3d data were both done via Open3d library. Code is supplied.
📂 Folder: 3d_reconstruction
Implements foreground/background segmentation techniques using classical and learning-based methods.
Useful for video analysis, surveillance, and motion tracking. As of now there is only patch-based approach, which is to be honest not that affective with an old web cam and bad illumination.
📌 Note: Requires the image of background without the actual subject in it (yeah its old tech).
📂 Folder: Backgroundsubtraction
Algorithms to stitch multiple images into a seamless panorama.
Includes feature matching, homography estimation, and blending. OpenCV's own stitching algorithm is pretty good. But i wanted to play with different approaches. As of now there are two approaches. First one is custom made which utilizes cumulative homography to both sides (not just left or right).
📂 Folder: Imagestitching
A collection of machine learning projects and experiments.
As of now it includes a single layered neural network that learns to classify inputs into two categories. Activation function :sigmoid loss function :mean squared error learning rate :0.01 size :2*1 epoch :1000
utilizes a custom made plotlib library to project points on a chart
📂 Folder: ML
Utility scripts and helper functions used across various projects.
May include image processing tools, data loaders, or visualization aids.
📂 Folder: Utils
Feel free to explore each folder for code, documentation, and results.
Updates and detailed descriptions coming soon!