Meetup Computer Vision Paris #20

👋 The Second Meetup of the 2021/2022 season will be on Wednesday, December 1st in physical !

As usual, there will be two talks on the programme. Each talk will be followed by a question and answer session with the speaker.

The program:

🌍 At Preligens, we use Deep Learning to automatize Satellite Images analysis: Object detection for strategic site surveillance, map creation for operations support, etc. When it comes to cartography, highly performing object detection or image segmentation models are not enough to create accurate and usable maps.

This is because a good map is not only accurate but also readable, which requires local and global coherency (e.g. buildings alignment) and polygon simplicity. How do we infuse these criteria in our predictions to create beautiful maps ?

🩺 Outpatient and ambulatory care is growing very rapidly in hospitals these days. Your home is tomorrow’s hospital room. But you need to be monitored when recovering from an operation at home, and Norbert is the right device for that - a simple white box loaded with sensors and
compute power will keep an eye on your key vitals without requiring you to do or wear anything. The same box will be used to monitor patients in post ICU rooms in the hospital, or to speed up the vital signs scanning process when checking in to a healthcare provider.

If you think further, contactless and passive health monitoring will enable large scale, continuous health monitoring with all demographics, and create a treasure trove of data to help detect issues very early on, by detecting slight changes in vital patterns. Detecting health issues before the first symptoms, health care providers will take care of those issues before they actually are visible. This is true preventive medicine. In your home or office.

At Norbert Health, we built a unique contactless vital signs sensing software-based method that relies on cutting-edge computer vision algorithms. We have developed and produced a device composed of three sensors - a standard visible camera, a long wave infrared imager, and a millimeter-wave radar - and an embedded computer board. We are able to check key vital signs: temperature, heart/breathing rate, blood oxygen level, pulse wave velocity, which are on their way to be medically certified. We are constantly developing new features to allow more vital signs acquisition, and smarter predictions such as diagnostic or alert.
Here, we will talk about our two core functionalities: body temperature estimation via RGB-infrared cameras calibration; heart rate measurement via minute detection of skin color variations. We will outline the algorithmic principles on which these functions rely, as well as the
technical challenges we have encountered. Last but not least, you will be able to interactively assess your own vitals using our device.

Norbert Health is an early-stage startup in MedTech, with offices in Paris and New York. We are backed by top tier VCs (most notably, Serena Capital and HCVC), and raised 9M so far. We are a small (<20) team composed of hardware, software, and machine learning/computer vision
engineers and of business/marketing experts. We have deployed more than 50 devices, have recorded more than 1 million vital sign scans, and are collaborating with hospitals and doctors to continually improve our product.

We hope to see many of you there!