![[Translate to Englisch:]](/fileadmin/user_upload/FCE/Stage/Body-perception-software-Stage.jpg "[Translate to Englisch:]")
Body perception software
Software to detect body properties and human pose enabling use cases that require exact understanding of the human body – in 3D.
Our solutions offer complete 3D body perception delivering details on joint and segment level as well as specific body key points. We extract body properties like height, weight or body shape as well as dynamic 3D Pose and offer interpretation on those. Our software solutions are hardware agnostic and can be deployed with different sensor types and compute platforms. We offer monocular 3D motion capture as well as multi sensor approaches.
If you have a need to understand the human body and integrate it to your application in real-time, come talk to us. We are helping customers globally to make their products and technologies more human, interactive and individual with a focus on automotive applications.
Our software is designed for embedded and edge computing and can be integrated to various solutions and architectures. We are robust in a variety of scenarios and noise factors. We bring in a lot of application expertise in human motion capture with many years of experience of data use. We are committed to help a world of tech understanding us as humans in the most naturalistic way.
How does it work?
Our cutting edge AI extracts a number of occupant features from camera images (like bounding boxes, keypoints, segmentations and more). Our 3D body pose AI is reconstructing a 3D avatar out of the image and features. We can use this digital representation of the human to extract all kind of body related information in your use case.
Which information can I get?
You get
- Presence: How many people are visible and what is their location
- Classification: Classification based on height and mass of occupants
- Body Pose: Location of all joints & segments
- Keypoint locations: 3D position like eyes, nose and others
- Center of mass for body and segments
- more application and use case specific data derived from the above
Which sensors can be used?
We can work on 2D images (RGB or IR) as well as 3D cameras (including depth information), for example Time of Flight cameras.
Is it 2D or 3D?
We deliver 3D pose information.
But, many 2D preprocessing steps are involved and can be made available if needed...
Which computer hardware can be used?
Our software is hardware agnostic and runs on Windows, Linux and any POSIX based system. We use machine learning algorithms thus benefit from GPUs or ML accelerators. We can run on high performance control units or small embedded devices. Examples are Orin, TDA4 or Ambarella CV5.
What is the accuracy?
Please contact us with more information on your use case, so we can provide you with information on our experiences and KPIs for those
Why work with Simi?
Simi has 30 years of expertise in human motion capture. Our roots are in biomechanics for sport and medical applications. Since 2019 we are transforming automotive OMS with our unique technology. We are developing our own technology and software products with highly motivated and skilled engineers combined with in depth understanding of our use cases.