X2R: Counter-'Counter-Reconnaissance'
Abstract
Degrading enemy surveillance while raising their uncertainty through deception is an overarching Force Design Imperative highlighted in the 2022 CNO (Chief of Naval Operations) Navigation Plan. The goal is to leverage deception measures such as electronic warfare, concealment, and maneuvers to control situations in contested battlespaces. However, it should be obvious to expect our adversaries to study and implement similar strategies in order to win the confrontation. In anticipation of our adversary's efforts in counter-surveillance, this project aimed to identify an effective approach that can defeat a certain counter-surveillance tactic, technique, or procedure(TTP). Specifically, we wished to apply an existing artificial intelligence/machine learning method to detect the presence of camouflaged objects. Within the limits of time and budget, we were able to study, implement, and modify the open-source Search Identification Network (SINet) model for testing and evaluation on visible images of animal and human subjects. Our experiments validated the usefulness of SINet and yielded some promising results. More surprisingly, the model was able to perform well even on targets that were absent from the training set, without additional cross-domain or transfer learning techniques. Our future research plan includes further assessment of the camouflaged object detection (COD) algorithm in the EO/IR/RF (electro-optics/Infrared/radio frequency) domain. Finally, rapid evaluation of the method with a drone in real time is being considered in a follow-on proposal.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 05, 2023
- Accession Number
- AD1215966
Entities
People
- Anthony Tai
- Matthew Compton
Organizations
- Naval Surface Warfare Center Crane Division