Recognition Pictures

Recognition Pictures With AI In The New Normal

Recognition Pictures have what it takes to search for the right image you need. Check out this post to find out more. 

Recognition Pictures With AI In The New Normal

When it comes to pictures, artificial intelligence has been there since the 1960s under several names: computer vision and image recognition. But, exactly, what is computer vision?

Your brain is a wonderful mechanism, even if you don’t recognize it. It can extract more information from a single image than we know what to make of. Take a look at the image below.

Barry is a cool canine. In Hawaii, he enjoys surfing.

If I asked you what’s in the image, you’d probably tell me it’s a dog on the beach on a bodyboard, wearing red sunglasses and a Hawaiian necklace made of artificial flowers.

Warning: this is a spoiler. The day when a computer can achieve this degree of precision and generality at the same time has not yet arrived. Fortunately for us — otherwise we’d be out of business — there are already certain real applications where computer vision is quite useful.

Tell Me What You See

So, what exactly do we teach computers? Simple: recognize, identify, and locate things with varying degrees of accuracy. Barry and his pal Ducky will demonstrate what I mean. To keep things simple, I’ll show you the four most common activities seen in real-world apps today:

Classification.\sTagging.\sDetection.

Segmentation.

Tagging and classification

Classification (left): we believe there is only a dog and no cat. Tagging (right): both a dog and a duck are present.

The simplest and most basic duty is to determine what is in an image and how certain we are about it, i.e., the probability percent in the two photos above. There are two major issues to consider:

What is your list of items to detect?

This is referred to as an ontology. It’s cats and dogs in the first image. To keep things (extremely) basic, you must first tell the algorithm what types of items it should recognize. And, like with all simple things, it’s more difficult than that. It is not usually necessary to mention all of the items. However, because this is an open area of research known as unsupervised learning, we will avoid it for the time being.

Is there more than one thing in the same image?

When there is just one object present at the same moment, we refer to this as categorization (left). Otherwise, it is referred to as tagging when many items are included in the same photograph (right).

Recognition Pictures: Detection and classification

Detection (left): we know which box Ducky and Barry are in in the image. Segmentation (right): we have data down to the pixel level.

After we’ve answered the What, the next question is: Where are the items we’re looking for? There are two methods to go about it:

Detection produces the rectangle, or bounding box, on the picture that contains the objects. It is susceptible to minor mistakes and imprecisions in location, but it is a highly durable technology.

Segmentation is taken a step further. We determine which, if any, objects each pixel, the most atomical element of information in a picture, belongs to. The result is an extremely exact map, although one needs a large amount of meticulously labeled data. When you have to do it for every pixel, it’s a time-consuming process, but it may provide remarkable results. This is one reason why use-cases in healthcare, particularly cancer diagnosis, are becoming increasingly common.

These were the four main components of computer vision. Instance identification, face key-point detection, action recognition, tracking, optical character recognition, image creation, style transfer, denoising, depth estimation, 3D reconstruction, motion estimation, optical flow, and so on are all available. You get the picture: there’s a lot to do!

Deep learning vs. traditional computer vision

“Any sufficiently sophisticated technology is indistinguishable from magic,” observed Arthur C. Clarke, author of 2001: A Space Odyssey. My take on this phrase is that you will never comprehend and accept something unless you explain how it works. This is especially true when it comes to artificial intelligence.

When you start peeling the onion, you discover it’s simply another technology with advantages and disadvantages. It should not frighten you any more than electricity does.

Leave a Comment

Your email address will not be published. Required fields are marked *