Разработка модели компьютерного зрения не всегда начинается с кода. Иногда больше времени уходит на подготовку данных...

Developing a computer vision model doesn't always start with code. Sometimes more time is spent preparing data, marking and testing hypotheses. Luna Line from VisionLabs combines these steps in one no-code platform: here you can load data, mark up images, train a model and check the result. The platform supports classification, detection and segmentation tasks. Moreover, to test hypotheses, it is not necessary to immediately collect a large sample: VisionLabs talks about scenarios for training classification models starting from 50 images. For an ML team, this can be a way to quickly move from an idea for a CV problem to a working model and understand whether the hypothesis is worth developing further. More about Luna Line