Luna Line will reduce the time to create neural networks and the cost of projects by 20–30%.
According to VisionLabs, the average project launch takes 4–6 months and requires a team of ML specialists. With the new Luna Line solution, the first version of the AI model can be created in 2–3 days and new versions can be launched just as quickly after retraining.
According to the developers, Luna Line allows you to train ML models using several dozen images. The user needs to upload at least 50 images to the system and mark them up. Next, the user clicks "Train" — this step can take from several minutes to hours, depending on the amount of data for training. After that, the neural network is ready to work.
The platform also provides visual-language models (Visual Language Model) that can work with both images and texts. They allow you to automatically mark up media files and analyze what is depicted on them.
The platform is available in the cloud, and can also be deployed on its own servers. During development, VisionLabs tested the Luna Line platform for identifying and classifying defects in metallurgical production, searching for bus stops with broken glass, and checking the application of identification numbers on cars.
