Many of the developers and normal people might be wondering how to make AI tools. Most of the people think that AI is very deep and that extensive knowledge is required to make such. While this is partly true since it is hard to code but many companies are trying to make it easier for the people and developers. Lobe, a company which specializes in this is trying to make the Artificial Intelligence much easier for the people. It has a simple UI with drag-and-drop features. Lobe is primarily aiming for deep learning algorithms for machine vision. For example, it can help users build a tool which recognizes the count of animals on the screen and so on. Moreover, it will enable the users to make it without typing a single bit of code.
How Lobe Works?
While we tested out Lobe, it turned out pretty simple as it says. The website is quite simple and easy to learn. Once a person learns the basics of Lobe, he can create its own deep learning algorithms. A person tried creating an algorithm which recognizes alphabets when shown on the webcam. He learned through the intro video and made it by himself in a short duration. The steps are mostly similar to that of other machine learning systems like collecting data, labeling and setting up a neural system.
For doing this, the user has to log in to the website – Lobe.ai. Once registered, the user can select from a list of pre-defined theme. There are limited options for now but Lobe said that they will be increasing it further on. The neural network is formed by the connection of boxes through a line like a flowchart. We can look inside each line and change the way they process the data. The website also lets the user export their model to other AI Tools for further enhancements.
Lobe is in beta mode so we see few patches in and around the website. The website overall looks slick regarding the background and other designs. Building an alphabet recognizer was quite easier but the tool can be used for many such applications. No matter how much amusing the tech is, it still has some major flaws associated with it. Some scientist thinks that making the work simpler will result in the creation of unstable algorithms. While Lobe has made it simpler for the users, some data scientist regard it that such systems come every year. They try to create such tools to simply work but they fail to understand that the harder part is to know what to write rather than just writing it.
Mike Matas, the co-founder of Lobe said that “Lobe is not trying to match the level of the tools like TensorFlow and PyTorch. It is made for amateurs for trying their ideas out. Many people have simple ideas and want to implement them and this tool will surely help them with it. If someone has ideas and can feed the algorithm with 1,000 samples, the algorithm can work efficiently. Overall, tools like Lobe can help people open up to Deep Learning and try their hands on it”. However, Lobe will not be a free tool as stated by Mike but the pricing is yet to be finalized.