Project Abstract

During the ongoing conflict in Ukraine, there is increasing evidence of the widespread usage of machine learning techniques to aid in warfare. While the ethics surrounding this are undeniably complex, it is now, and in the future, an undisputable fact of warfare. Computer vision has, of course, been utilised in weapons systems for arguably decades, with the Storm Shadow cruise missile being one obvious example; however, these have always been products of high-end defence manufacturers with the associated high-end price tags. Increasingly, the Ukrainian forces are relying upon homegrown systems assembled from commercial off the shelf (COTS) components to service targets which would traditionally have been the domain of prohibitively expensive systems from Western manufacturers. One of the obstacles these solutions frequently encounter is the use of electronic warfare to deny either their means of control or of navigation. That is why this project aims to explore the extent to which it is possible to run computer vision algorithms on resource constrained devices (such as Raspberry Pi 4), which are readily available on the commercial market and could therefore be reasonably incorporated into an affordable solution, with the aim of aiding terminal guidance when within a hostile EW environment. To this end, a solution ?�� initially focussed on identifying aircraft on the ground – was implemented in Python using the OpenCV library, which was then run on actual hardware to determine whether the performance is adequate. The initial findings are that the concept is indeed valid, but that it might prove ideal to use slightly more powerful hardware in order to run more resource intensive models. In conclusion, this is a capability which is as frightening as it is impressive, and the democratisation of such capabilities is a threat which must be considered by all parties into the future.

Keywords: Computer Vision, Object Detection,

 Conference Details

Session: Poster Session B at Poster Stand 105

Location: Sir Stanley Clarke Auditorium at Wednesday 8th 09:00 – 12:30

Markers: Gregory Cheng, Scott Yang Yang

Course: BSc Computer Science, 3rd Year

Future Plans: I have a job lined-up