This repository contians a list of notebooks that sysytemaically implements the core concepts of computer vision. The notebooks starts from basic concepts simple image processing techniques and edge detection techniques and work their way up to more complex algorithms and technologies like U-net (for image segmentation)and GANs(for image generation)
The following table give the list of notebooks alongside a simple desciption to each of them
| Notebook | Description |
|---|---|
| Lane detection with houghs transform | Uses houghs transform to detect lines in edge map to find the lanes |
| Histogram equilisation, gamma corretion contrast streting | Performs some basic image enhancement techniques to improve the visual appearence of the image |
| Edge cdetection | Performs edge detection with sobel operator, prewitt operator and canny edge detection. |
| Dog breed classification | Employs a pretrained model to perform dog breed classification without any finetuning |
| Pretrained models | notebook trains and finetunes 3 different deep convolutional networks on cat vs dog data set and study the results |
| Image segmentation on FCN vs U net | performs semantic segmentation on VOC dataset using FCN and U-net architectures. |
| GAN using MNIST dataset | We build a GAN network to generate images of handwritten digits using a custom GAN model. |