Modern Computer Vision with OpenCV 2026

In this lecture, we will explore how to read images, videos, and live webcam feeds using OpenCV.

In this lecture, we will learn some basic OpenCV functions, including how to:

• Convert an image to grayscale

• Blur an image

• Detect edges in an image

• Perform dilation and erosion on images

• Crop and resize an image

• Draw shapes on images (e.g., lines, rectangles, circles) and add text to images

In this lecture, we will work with an image of a card and focus on isolating the King of Spades. Our goal is to flatten the card by applying a perspective warp using four defined corner points.

Key Steps:

1. Identify the 4 corner points of the card.

2. Define these points as a NumPy array of floats.

3. Use these points to apply the Warp Perspective transformation.

In this lecture, we will explore how to join or stack images together both horizontally and vertically using np.vstack and np.hstack.

Key Points to Consider:

1. Image Size: Images must be of the same size to be stacked.

2. Number of Channels: The stacking functions will only work if all images have the same number of channels.

In this lecture, we will learn how to detect a specific color in an image using OpenCV. As an example, we will focus on detecting the orange color of a car in the given image.

In this lecture, we will:

1. Detect and draw contours around each shape in the image.

2. Identify the corner points of each shape using the contours.

3. Draw bounding boxes around each detected shape based on these corner points.

4. Classify the shapes (Circle, Square, or Rectangle) by analyzing the number of corner points.

In this lecture, we will learn how to build a Document Scanner using OpenCV.

In this lecture, we will present a demo of the Streamlit application that we will be building in this section. The application demonstrates how to perform math using hand gestures.

In this lecture, we will learn how to use the MediaPipe library for hand tracking.  We will detect and track 20 different landmarks on human hands

Use your hand to create a drawing, which will be processed by an AI model to solve math problem. You can also ask the AI model questions about the drawing.

Learn how to create a real-time hand gesture-controlled spin wheel using the OpenCV and MediaPipe libraries.

In this video, we will learn how to estimate real distance to objects with Depth Pro and YOLO11 Model.

This lecture showcases a demo of the Tennis Analysis System that we will build from scratch. The project involves detecting and tracking players and the tennis ball throughout the entire video. You will learn how to detect the tennis court keypoints to determine player positions relative to the court. Additionally, the project covers creating a mini court and implementing player and ball movements within it. Finally, you will also learn how to measure player speed and shot speed.

In this video tutorial, you will learn how to train/ fine-tune the YOLOv5 model for tennis ball detection.

In this video tutorial, you will learn how to train/ fine-tune a Convolutional Neural Network (CNN) model to detect tennis court keypoints.

In this video tutorial, we will implement player detection and tracking using YOLO11.

In this video tutorial, we will implement tennis ball detection and tracking with YOLO11.

In this video tutorial, you will learn how to detect tennis court keypoints.

In some frames, the tennis ball is not detected at all. We can use the interpolate function in pandas to fill in the missing detections.

In the output video, we observe that many people have been detected. However, we only want to detect the two players on the field. To achieve this, we can select the players closest to the court and use the court keypoints to determine which players are nearest to it.

In this video tutorial, you will learn how to create a mini court.

In this video tutorial, you will learn how to identify the frame numbers when the tennis ball is hit.

In this video tutorial, you will learn how to implement player and ball movement within the mini court.

In this video tutorial, you will learn how to calculate player and shot speeds for performance analysis.

In this video, you’ll learn how to build a Pool Shot Predictor that uses geometry and vector projection to determine if the ball will go into a pocket. This project incorporates various image processing techniques, including thresholding, line detection, and contour analysis.

In this video tutorial, we will be creating a people counter that counts people entering and exiting using Ultralytics YOLO11.

In this lecture, we will build a Football Analysis System from scratch. This project focuses on detecting and tracking players, referees, and the football throughout the entire video using YOLO.

YOLOE released by researchers at Tsinghua University: Real-time seeing anything with zero-shot performance. YOLOE can do both object detection and image segmentation and supports different prompt types, including text prompts, visual inputs, or even no prompt at all.

?YOLOE results have been compared with YOLO-Worldv2, which supports arbitrary text prompts too.

Key highlights:

✅ Zero-shot performance: It can find and recognize new objects, even if it hasn’t seen them before.

✅ Training time: Based on a research paper, It reaches the same accuracy as YOLO-Worldv2 but in one-third of the time on the LVIS dataset.

✅ Inference time: Its prediction speed is slightly better in comparison to YOLO-Worldv2.

✅ Pretrained embeddings: It uses a stored Apple MobileCLIP text encoder to keep training fast.

In this video, you will learn how to perform object detection on images and videos using RF-DETR

In this video, we will explore how to train the RF-DETR object detection transformer on a custom dataset for pothole detection.