NIH/FDA Computer Vision & Embedded Systems for Automated Monitoring of Research Animals

An ORISE fellow will be instrumental to advancing the progress on our collaborative project for automated tracking and behavior detection and analysis of research animals (rodents). The fellowship will offer an exciting opportunity for the fellow to gain in-depth knowledge in computer vision, deep learning, and embedded systems.

The research involves tailoring the latest algorithms developed by leaders in the field for object detection and pose estimation to track and potentially uniquely identify (in case of multi-animal setups) multiple animals. The research also involves adapting behavior detection methods to profile general behaviors (e.g., walking, running, etc.) of the animals as well as behaviors of special interest to different investigators. One critical requirement of these systems is real-time processing. Achieving real-time processing will be sought by optimizing algorithms as well as implementing the algorithms on edge devices (e.g., Nvidia Jetson devices).

The research will allow the fellow to gain and solidify skills in in computer vision, deep learning packages such as PyTorch, and embedded systems programming. This is a collaborative project with the NIH Institute of Biomedical Imaging and Bioengineering. The research effort will be summarized in conference and journal publications for which the ORISE fellow will get authorship recognition.