Concept ‘Benchmark Data Consortium for Explosive Ordnance Detection Algorithms’ (6 October 2023)

Since it began work to create computer vision detection algorithms for explosive ordnance objects in 2022, T4T and VFRAME have developed a robust dataset of benchmark data and video against which to test our algorithms. But we recognise that to truly validate our tools they must be tested against a wider pool of data that has not been used to create the algorithms. The same goes for all vendors that claim high-performance detection in this space. For this reason, we are proposing the establishment of a Benchmark Data Consortium for Explosive Ordnance Detection Systems, whose role will be to provide independent validation and certification of algorithm accuracy against a benchmark dataset generated by contributions from actors developing new tools, as well as EOD centres and other approved actors.

We invite other vendors, whether NGOs or for-profit companies, to participate in the Consortium, and to agree that independent verification is the only way to disprove false claims and misinformation about algorithm accuracy. With the lives of emergency service and mine action personnel on the line, the urgency to establish such a system is real.