ML and AI

Object Detection with Cloud AutoML from Google

Object detection and recognition algorithm highlights different types of masks on streetwalkers

Object detection is a machine learning technology that can benefit many products in different business domains. For example, it is particularly useful when it comes to score and text recognition systems that we have discussed in one of our previous articles. It was only a matter of time that the growing demand for this kind of solution would urge the industry leaders to create an easy-to-use packaged tool. Today SOLVVE would like to talk about Cloud AutoML from Google and what our machine learning experts were able to achieve using it.

What’s Object Detection and How it Works?

Simply put, object detection is a part of computer vision and computer processing technology that can tell if a certain type of object, living being included, are present in a picture or video. For example, surveillance systems rely on facial recognition technology to provide advanced security features.

Object detection should not be confused with object recognition, although oftentimes these terms are used interchangeably. Object recognition can not only tell if something is in the picture, it can also tell more about the properties of an object. For example, if there is a flying object, then is it an airplane or a kite. 

These two technologies combined provide a powerful tool to create sophisticated products and solutions. For example, to recognise and understand different kinds of activities, even such complicated ones as human behavior to prevent crimes.

There are many different systems that provide object detection and recognition tools of different levels of complexity. And Cloud AutoML is one of them. Let’s take a look at what it can do.

Why Choose Cloud AutoML?

One may ask, how is it different from other object detection tools, and why should your business opt to use Cloud AutoML instead of other existing instrument.

Firstly, Cloud AutoML does not require any coding. That will significantly cut down development and training times, and bring your desired features to life faster. That also translates into lesser financial input since you do not have to cover extra manhours that go into the coding.

Secondly, Cloud AutoML is valuable for its object localization capabilities that allow not only to find multiple objects in the image, but also to provide information about them as shown in the picture below. Depending on the needs of your business you can train your algorithms to detect, identify and classify a vast majority of objects with high accuracy. 

Thirdly, Cloud AutoML does not stop at training, but allows to easily integrate your custom models with devices, including those running on Android or iOS. It means that if you need an object detection or recognition feature for your app, Cloud AutoML makes it faster and cheaper than ever before.

I Would Like to Have This Feature in my Product

Even though Google describes its product as a tool to train “high-quality custom machine learning models with minimal effort and machine learning expertise”, using it does require a skilled ML-engineer to deliver efficient solutions fast. 

The machine learning department at SOLVVE already has hands-on experience with this instrument and can help you plan, develop and deploy object recognition features for your product or application. If you have any questions or ideas about using machine learning for object detection and recognition in your business, do not hesitate to contact us. Let us make it happen!