Data Science R&D center Winstars.AI

DESCRIPTION

Virtual replacement takes 4 steps:

• Indoor scene parsing
• Extracting floor plane
• Texture transformation
• Insertion improvements (contours and colors)

TECHNICAL DESCRIPTION

These steps are possible because of the DeepLab Xception model trained on the ADE20K dataset, which contains 150 classes of objects. All manipulations predicted by model flooring were done with Open CV. Also, with this library, floor texture automatically rotates and crops to the size of the original image of the floor.

BENEFITS

The program is easy to use: all you need to do is upload a real photo of the room, add the desired texture, and specify the inclination angle. Next - rejoice in the finished result)

Thanks to this project, designers can easier depict an interior and workers – calculate materials and laying area.

DESCRIPTION

According to the International Diabetes Federation, over 537 million people are living with diabetes globally, which is growing at an alarming rate. Most of these people will develop some form of retinopathy, causing reduced vision or, in severe cases, blindness.

We created a cloud-based medical solution to detect diabetic retinopathy using photographic images of the eye fundus through the ML\AI algorithm.

Photographs can be taken anywhere – from optic retailers to pharmacies – by non-medical staff to be further uploaded to the system for diagnosis. CheckEye will make early detection of diabetic retinopathy and retinopathy prevention easily accessible.

WHO BENEFITS?

Patients:

• reduced waiting time;
• better cost efficiency;
• timely treatment;
• better vision for life.

Eye health professionals:

• more quality time with patients;
• better performance results;
• accurate diagnosis.

Public health systems & insurance companies:

• better access to screening;
• more accurate screening results;
• less financial burden;
• less admin work for medical personnel.