What is the different between detection and recognition

Hi everyone,
I’m learning third course about CNN in Deep learning specialization. After learned 4th week about object detection, i wondered what was the main different between recognition and detection ? When we do detection, it also has classification in that. So what is the biggest different between that ?

An informal answer:
To me, detection seems to be like a Yes/No question. “Yes, there is a cat in this picture”.

Recognition seems more like “There are five cats in this picture, and their species are…”.


By the time we get to W4 of DLS C4 (ConvNets), we’ve seen a number of different levels of sophistication in terms of object classification and object recognition and localization within an image. In DLS C1, we learned how to build a basic image classifier that (as Tom says) can recognize “is there a cat in this picture or not”. Then in DLS C4, we learn about more sophisticated algorithms like YOLO and U-Net for Image Segmentation that can not only identify multiple different objects within an image, but can also draw bounding boxes around them (in the case of YOLO) or label the individual pixels that are the components of that object (as in U-Net Image Segmentation) and also classify the type of object (car, pedestrian, traffic sign, road surface …).

1 Like