Reference Number: ARL-SRP-00091
Project Description
Active visual search (AVS) refers to the task of searching for a set of objects or regions of interest by changing pan, tilt, zoom (PTZ), and focal length of a PTZ camera to better recognize each object of interest. This is an activity done by people all the time, especially when they have access to binoculars yet this remains a difficult task for AI algorithms. The focus of the summer project will be to introduce new capabilities to an existing active visual search system. The focus will be either to improve the scene understanding of the system beyond object recognition (integrating depth estimation, object tracking, or visual segmentation models). Alternatively, based on interest the project may focus on extending the existing system to control PTZ security cameras over ROS2 along with using a visual language model to execute known human visual search methods.
Required skill set: Willingness to learn new things, experience with python or C++.
Requested Skill Sets: Experience machine learning, computer vision, ROS2, OpenCV, or prompt engineering with visual language models is beneficial but not required. Some basic electronic hardware experience can be helpful if you wish to use a physical PTZ camera.
Technical Skills
- Advanced Sensing
- Artificial Intelligence/Machine Learning
- Computer Science
- Data Science
- Electrical Engineering
- Electronics
- Mechanical Engineering
