Although computer vision has been extremely successful in employing deep neural networks for many visual perception applications, its use is still hindered when it comes to practical medical applications. This shortcoming is due to the factors include lack sample data, the cost of data gathering and labeling, patient privacy and safety and accessibility, and unobtrusiveness of the data collection/monitoring technologies. In‑bed pose estimation has practical value in medical and healthcare applications, yet mainly relies on expensive pressure mapping (PM) solutions with limited granularity which roar up to thousands of dollars.  

Needed is a method of monitoring patient movement and comfort for situations such as the avoidance of secondary medical issues (bed sores, pressure sores) and data collection for sleep patterns.


Northeastern inventors have focused on the critical healthcare application of in-bed patient monitoring using a physics-inspired vision-based approach, called Under the Cover Imaging via Thermal Diffusion (UCITD), It accurately estimates in-bed poses with high granularity under natural sleeping conditions and it works under natural conditions (e.g. full darkness and heavy occlusion), and is contact-less and unobtrusive, medically-safe (radiation free).

A key contribution comes from the accurately annotated SLP dataset.


  • Can monitor under the cover or in full darkness 

  • Contactless approach preserves natural setting including bed covering

  • Superior accuracy and resolution in human pose estimation yet only a fraction of price  and size 


  • Sleeping behavior studies 

  • Patient activity monitoring in hospital

  • Pressure ulcer studies: early detection and prevention

  • Healthcare research that requires in-bed human pose information over time  


Patent Information:
For Information, Contact:
Myron Kassaraba
Director of Commercialization
Northeastern University
Sarah Ostadabbas
Shuangjun Liu
computer vision and deep learning
In-bed pose estimation
Sleeping behavior monitoring