An image classifier answers one question about a whole picture: what is in it? Object detection goes further. It finds every object in the image, draws a bounding box around each one, and labels it with a class and a confidence score. That is what powers camera apps, retail analytics, driver assistance and many other products.
The lessons start by separating classification, detection and segmentation. Next you learn how the YOLO (You Only Look Once) algorithm detects objects in a single pass, then run a pretrained YOLO model on your own images. The last lesson trains YOLOv8 on a custom dataset, so your app can detect the objects you care about.
Watch the lessons in order and code along in Google Colab or a Jupyter notebook. A GPU runtime in Colab makes training much faster. Once you have a working detector, wrap it in a small app that takes an image or webcam frame and draws the boxes on it.
A simple worked example of how image classification, object detection and image segmentation differ, and when to use each.
Video 2minA plain-language look at how YOLO (You Only Look Once) detects objects in one pass: the grid, bounding boxes, confidence scores and non-max suppression.
Video 16minUse YOLO v4 with pretrained weights trained on the COCO dataset to detect object boundaries in an image, including the Darknet setup.
Video 15minPrepare and annotate your own dataset, configure YOLOv8, train and fine-tune the model, then test it with inference on new images.
Video 15min