Posted: 12-10-2024
Title: Deepfake Detection
Location: ARL West
Description: Deepfake – an emerging AI digital manipulation technology is being increasingly weaponized, posing a significant threat to our society and national security. The original concept of deepfake or AI-synthesized hyper-realistic images or videos, has been decried primarily in connection with involuntary depictions of people.
Recently, significant concerns have been raised about a far more nefarious threat, Deepfake geospatial data (i.e., satellite images, maps, digital terrain models, etc.), which drives deepfakes to another level. Geospatial data plays a pivotal role in Army mission planning and operations. Imagine a scenario in which an intelligent analyst or mission planning software is fooled by fake satellite image or map that shows a non-existent bridge in marching route, or a terrain model of fake road obstacles being transmitted to an autonomous vehicle to mislead its navigation system. Deepfakes are increasingly used to manipulate scenes and pixels/objects to create artifacts on geospatial data for malicious purposes.
This project aims to develop novel techniques and solutions for detecting and defending against deepfake attacks on geospatial data. It will focus on theoretical research and practical algorithms, which enables deepfakes of geospatial data to be detected and defended respecting the functional and physical properties of real scenes.
We have developed a breakthrough method for detection of deepfake face images to support mission-critical data integrity protection. This technique achieves a lightweight, low training complexity and high-performance deepfake face detection. We will enhance and extend this theory and framework for the detection and recognition of deepfake geospatial data including satellite images, maps, and digital terrain data.
Anticipated outcomes include new theory and algorithm developments leading to publications in scientific forums and real-world utility and software for Army evaluations.
Requirements: Dedicated and hardworking individual; Experience or coursework related to machine learning, signal processing, computer vision; Strong programming skills.
Level: Multiple Classifications
Required Major(s): Computer science, Electrical Engineering, Applied math
Find a Volunteering Opportunity
Visit our Program Volunteers page for a tool to find the best opportunity for you.
eCYBERMISSION Grants
The eCYBERMISSION Mini-Grant is intended to support teachers/program leaders as they implement eCYBERMISSION with their teams. Educators (formal and informal) of students in grades 6-9 are encouraged to apply. Special consideration is given to Title 1 schools and to those with underserved/ under-represented populations.