A repository containing all the projects assigned from the "Computer Vision" course in ECE DUTh.
The main tool used is OpenCV.
Text Division and Segmentation Project.
- Median filter for noise removal from images.
- Region Segmentation with dilation and closing techniques.
- Design of bounding box with connected components function.
- Word and region counting as results.
| Image Original Data |
Resulted Image |
 |
 |
Panoramic Image Stitching
- Key features detection using Scale Invariant Feature Transform (SIFT) algorithm.
- Key features detection using Speeded up Robust Feature (SURF) algorithm.
- Cross-checking features according to Manhattan distance.
- Homography for image transforms.
| Original Photos for the Panorama's creation: |
 |
| Crosschecking Result (SIFT) |
Crosschecking Result (SURF) |
 |
 |
Image Classification using Bag of Visual Words Model
- Using part of Caltech-256 Dataset.
- Creating Histograms and Descriptors.
- Creating Dictionaries.
- Extraction of local features.
- Training a BoVW model using K-Means.
- Classification with Support Vector Machines (One versus all).
- Classification with K-Nearest-Neighbors.
Image Classification using Convolutional Neural Networks (CNN)
- Using part of Caltech-256 Dataset.
- Different Deep Neural Network architectures
- Data Augmentations
- Pre-trained architectural model, VGG.
Contributor: Associate Professor Lazaros Tsochatzidis year 2022.