Computer Vision with Python

Load an image with the OpenCV computer vision library, display it in a window, wait for a key press, then close the window, and handle 64-bit and 32-bit setups.

Learn how to convert color images to grayscale in OpenCV using both direct grayscale loading and cvtColor, and display color and grayscale results side by side.

Import and use an image module to fetch metadata with shape, then print the image's height, width, and color depth.

Explore how image pixels encode color with red, green, and blue values from 0 to 255, and how changing pixel coordinates and channels can produce white when all reach 255.

learn to crop images with opencv by specifying start and end points in brackets and adjusting x, y, and size to target image regions.

This lecture demonstrates color-mean features in computer vision by computing the average red, green, and blue values of an image and converting them to integers.

Explore grayscale histograms to describe images by dividing color space into regions, forming feature vectors that quantify dark, middle, and bright areas; learn when color helps or fails.

Extract texture-based features by applying Local Binary Patterns on image blocks, comparing each center pixel to its surrounding neighbors to form 8-bit codes and a final feature vector.

Install the local binary pattern package, compute the LBP across the image with eight points and radius three, and normalize histogram of pattern frequencies to form a texture feature vector.

Learn how a reverse image search engine uses feature vectors and Euclidean distance to compare a query image with a large image database and find the most similar images.

Build an image retrieval system that searches for similar images using feature vectors and distances, using local binary pattern texture features and a database to rank results by distance.

Template matching locates image areas that match a given template or patch. The computer automatically finds occurrences but lacks scale and rotation invariance.

Build an object detection app with template matching in grayscale, loading a search image and a pen template, and drawing rectangles for matches; it isn’t invariant to scale or rotation.

Learn to detect faces in images using Haar cascade in Python, convert images to grayscale, apply the cascade classifier, and draw rectangles around detected faces.

This lecture explains how Haar features are extracted by subtracting sums of pixels under white and black rectangles, used in a cascade classifier that slides a window to label images.

Explore how keypoint features identify distinctive points to recognize or compare objects under rotation and occlusion, while highlighting diverse state-of-the-art algorithms.