Exercises related to computer vision

• YOLO - Implementation of the state-of-art real time object detection algorithm You Only Look Once using a pretrained model.
• Attention - An implementation of attention in isolation from a larger model. Paper: Effective approaches to attention-based neural machine translation.
• Optical-Flow - Implementation of optical flow algorithm, it tracks objects by looking at where the same points have moved from one image frame to the next.
• GOTURN - Generic Object Tracking Using Regression Networks. Program to keep track of an object in a video sequence.
• CNN-Layer-Visualizer - Program to visualize the output of a convolutional layer, an activated convolutional layer, or a pooling layer.
• ORB: Oriented Fast and Rotated Brief algorithm implementation using OpenCV to perfom object detection.
• HOG: Histogram of Oriented Gradients algorithm explained and implemented using OpenCV
• K-means: Perfomed image segmentation using K-means clustering.
• Clothing-classifier: Clothing classifier using FashionMNIST dataset, 90% accuracy, model from scratch.
• Face-detection: Face detection using OpenCV, especifically performed using a haarcascade detector.
• Circle-detection: A dectection program using The Hough Transform to detect consistent shapes, in this case circles.
• Contour-detection: Perfomed image segmentation by detecting contours.
• Corner-detection: Detect corners (interection of two edges where gradient is high on all directions) using Harris Corner Detector.
• Day-or-night: A day and night image classifier that uses the average brightness of an image as a threshold to perform the classification.