One popular family of object recognition models is referred to collectively as YOLO "You Only Look Once". It's developed by Joseph Redmon.
This model approaches a single neural network trained end to end that takes photograph/video as input and predicts bounding boxes and labels for bounding boxes directly. With YOLO, you can get almost immediate results very easily.
Prior detection systems repurpose classifiers or localizers to perform detection. They apply the model to an image at multiple locations and scales. High-scoring regions of the image are considered detections.
YOLO uses a totally different approach. It applies a single neural network to the full image. This network divides the image into regions and predicts bounding boxes and probabilities for each region. These bounding boxes are weighted by the predicted probabilities.
Click the pulsating plus-signs to see explanations.
Tähän hotspot- komponentti kuvineen
Summary of Predictions made by YOLO Model. Source: You Only Look Once: Unified, Real-Time Object Detection
Cool, huh?
You will be able to do the same with your own videos.
One thing that we will be needing during the upcoming lessons is Darknet. The name might make it sound quite ominous, but Darknet is simply an open-source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation. It can be used for many purposes, and one of those happens to be YOLO. It has also been created by the same person, Joseph Redmon.
You can find more information about Darknet and YOLO from their official websites, or you can check the following PDF, made by the creator of YOLO.
Tähän lataa PDF- komponentti