Computer Vision application for cell recognition

The processing pipeline consists of two stages based on artificial neural networks. It detects microorganisms and then segments their bodies to determine their surface on the image. To achieve such a precise segmentation, it's necessary to overexpose the image on a molecular level with support from other fields such as physics, mathematics, technology and more.

Flyps achieved a 95% recognition level accuracy and increased the high-resolution photo analyse time from 30 sec to 2 sec. Problems we had to overcome with our solutions:

• Recognition of the diversity of appearance of organisms from the same group
• Poor quality and focus of photographs taken by the microscope
• The delineation between cells that are either clustered or just dividing
• The similarity in appearance of different organisms on the microscopic image.

‍