Even if Artificial Intelligence is becoming more and more popular, it may seem an abstract and unrealistic concept in our daily life; however, we are constantly finding new ways to implement this technology in our ordinary life.
Artificial intelligence is precisely in charge of looking up solutions for any problem that may exist in the world - and if not, a solution could be created - precisely developing technology that emulates and enhances human capabilities, such as calculating, memorizing or learning.
Usually, we associate this type of technology with Machine learning, which consists of training an algorithm based on patterns that are not defined by us, and which " learn " automatically from the data.
We will soon explain in more detail how Machine Learning works and other interesting subjects in the blog. , stay tuned to our publications to find out more!
Check out our demos!
Komorebi wants to help you to experience practical applications of artificial intelligence through our short demos, in which we show you in brief how algorithms we build can perform some features of AI.
Our first demo is Art-explorer; this application has a collection of about 86,000 paintings, in which you can do a quick search through simple text, understanding natural language.
Art-explorer will translate the concept or description we gave and return the most appropriate pieces of art from the collection. We will have in seconds between 1 and 5 examples close to our description.
Click here to try the demo: https://art-explorer.komorebi.ai/
Users can find from concrete concepts, for example: "A painting of a woman in a red dress", the application will let visualize the following examples:
Also the application will allows user to search in multiple languages (Spanish, English, French, Portuguese), releasing a few variations of style in the outputs, returning to the example of "A painting of a woman in a red dress", these would be the first examples for each language:
In this demo, you can experience a semantic search technology based on zero-shot recognition, which allows you to search for some more metaphorical concepts, for example: "Flying in pink love", where art-explorer will give you approximate results to the context you are describing:
In this demo, you can experience a technology based on zero-shot semantic search, which allows the recognition and classification of images associated with a concept.
Often algorithms need a large amount of training data, which is not available in practical scenarios. In this demo, we present a model that is not trained on the specific question; however, it uses algorithms that generate results based on similarity to the searched concept.
How does it compare to Google image search?
Regardless of our small database, the main differences start to appear when providing complex but vague queries, where thanks to the development of machine learning Art-explorer will get closer results. For example if we use the description "a cubist painting with some wood in it".
These are the first four results that we will get in Google: (only one satisfies the query).
And this will be what art-explorer shows:
The models developed with specific uses are applicable to multiple fields, if you have more ideas of #ai apps that can be useful to your business; do not hesitate to consult us at firstname.lastname@example.org, we will be happy to listen to you.