abdulazeem / computer-vision-udacity

Computer Vision Nanodegree Udacity

Computer Vision Nanodegree Udacity

Here you can find several small (and slightly larger) subprojects dedicated to the well-known Computer Vision approaches and algorithms. These subprojects are developed as part of the Udacity Computer Vision Nanodegree Program.
Several projects are devoted to Deep Learning Arcitecture of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN):

• FashionMNIST Convolution Network
• YOLOv3 notebook implementation
• CharacterLevel RNN with LSTM
• Facial Key Points CNN (Project 1)
• Image Captioning CNN+RNN (Project 2)

Projects

Your can view and launch the following Python projects:

Fast detector of facial keypoints (ORB)

Hough Transform: Detect Lines and Circles

Steps of Canny Edge Detector

Binary Classifier for Day-Night Images

Edge Detection with Sobel Filters

Find contours and their directions

AdaBoost frontal face detector

K-means clustering and Masking

FashionMNIST Convolution Network

YOLOv3 notebook implementation

High-Pass and Low-Pass Filters

Comparison of Color Models

CharacterLevel RNN with LSTM

Project 1: Facial Key Points CNN

Project 2: Image Captioning CNN+RNN

Project 3: Graph-SLAM algorithm

Used OpenCV functions

• cv2.imread
• cv2.cvtColor
• cv2.ORB_create
• cv2.BFMatcher
• cv2.drawMatches
• cv2.GaussianBlur
• cv2.circle
• cv2.HoughCircles
• cv2.Canny
• cv2.HoughLinesP
• cv2.line
• cv2.inRange
• cv2.bitwise_or
• cv2.bitwise_and
• cv2.resize
• cv2.filter2D
• cv2.threshold
• cv2.findContours
• cv2.drawContours
• cv2.fitEllipse
• cv2.boundingRect
• cv2.rectangle
• cv2.CascadeClassifier
• cv2.kmeans

Environment

Make sure that Python and libraries numpy, cv2, torch, matplotlab, os, utils, darknet are installed.
The additional info about every project can be found in the README file related to the specific project.

Credit

I liked the course "Computer Vision Nanodegree".
Many thanks to Udacity for the high learning standards.
Most of the code in this repository is based on Udacity code.