Movie search service
With the usage of neural networks and algorithms, the service will help you find a movie by its description or recommend a similar film to the one you liked.
Have you ever been in a situation when attempting to find a movie title, but you only remember its storyline, some main events, or actors? Or have you ever wanted to get a recommendation of a movie of the same genre or a similar theme, but online services do not allow you to customize your search enough?
Movie search service is perfect for such cases. With the usage of neural networks and algorithms, the service will help you find a movie by its description or recommend a similar film to the one you liked.
The system contains a parser that collects information from the IMDB website.
Milvus allows to store data in vector format and retrieve records from the database using different algorithms. In this case, the L2 norm, or Euclidean distance, was used to extract the most similar films for the input vector of the user’s request.
The service has two search query modes: text query and a combination of text keywords. In both cases, the text query is converted into a vector by the BERT model and the most suitable results are extracted from the Milvus. If keywords are used as well, then the names/surnames of the actors or directors are extracted from the text query using Stanza’s NER models and after that, the previously extracted movies are sorted in the order of the keywords found among the description/cast.
The web version of the service is written with Django.
Complex solution for oil consumers and suppliers
The platform for consumers and suppliers of oil and petrochemicals to set a feasible connection between them.
Virtual try-on tool for makeup products
The system consists of a face detection and segmentation model and an algorithm that allows recoloring objects without losing their original texture.
Online sign language interpreter
AI algorithm that converts video of a person using sign language into a text transcript