Supporting video link: https://www.bilibili.com/video/BV1ZL4y1p7Cz/
First of all, everyone needs to understand what target detection is. Only in this way can we determine whether our task is target detection, so as to ensure that the following content can help you solve the problem.
Object detection, Object Detection. When I first saw the word detection, I thought it was just to detect the position of the target in the picture, but later I found out that I was too young.
Detection not only needs to find the position of the target of interest on the picture, but also recognizes what category the target is. As this image shows, object detection finds the location of three classes of objects of interest in the image: dogs, bicycles, and cars. Not only are the locations of these objects found, but the class of objects at the detected locations is also identified.
The application scenarios of target detection are very wide, mainly to grasp the word target. The definition of the target is defined according to the scenario of our application. For example, in face detection scenarios, we can use faces as targets. In the text detection scene, we can treat the text as a scene. More scientific research now focuses on object detection in natural scenes. The detected objects are mainly common object types in life. The characteristics of this type of detection problem are that there are many types of objects, and the location of objects is more complex and diverse.
Therefore, if the target detection model can achieve good results in natural scenes, it should be able to achieve good or even better results in most application scenarios. All this requires us to slightly modify the network model, choose a suitable application scenario or create a corresponding data set, and then we can obtain excellent detection results.
This article is published by OpenWrite, a multi-post platform for blogging !