UM2, Milestone responded with clarification on Milestone integration, I double checked all cameras and Hanwha, Hikvision, and Dahua integrated face detection events while Uniview and Vivotek did not, I updated the chart and text to reflect this.
There are these tests recently published from NIST in which only Dahua is referenced. It would be interesting to know which algroithms sets and code can be integrated onto a camera and/or VMS platform. Some interesting results in the NIST report generally as regards performance with masks (pre-COVID 19) and to look at what improvements might occur as manufacturers address this challenge and as they tackle mask detection as well. Do you know if any of these vendors (apart from Dahua) intend to participate in ongoing NIST testing?
Have you thought about getting reflectacles as opposed to sunglasses and seeing how these do when pitted against anti-surveillance techniques?
Hanwha also has face detection with the face overlay in their PVMs and TNB-6030 PVM camera. You can select the color of the overlay as well as text on screen such as "RECORDING IN PROGRESS" with the option to blink the text as well to get attention. This helps deter offenders in retail stores.
So is this similar to the LPR / LPC issue where a lot of cameras can capture a license plate but very few can recognize them? You are detecting faces in this test but what can you do with this information now? Is the idea of face detection for the use in face recognition?
Ethan, I would be interested to find out if this basic face detection can translate into a searchable criteria in the various recording scenarios that might be used whether stand-alone NVR or separate VMS/CMS. If so that's one more way to filter out spurious events when searching.