Airport Screening Re-Identification Video Analytics Program Examined (TSA CLASP)
By Zach Segal, Published Jan 21, 2021, 10:50am EST
(Info+)
The US TSA's CLASP program is targetting re-identification and luggage-person association from ceiling-mounted cameras to enable multiple screening levels in one lane, one of the more ambitious Western video analytics programs we have seen.
IPVM explores the program and how the analytics could be adopted to other uses including feedback from the Department of Homeland Security (DHS) researchers behind CLASP to IPVM.
I wonder what analytic technology they will use to identify a human and search for a re-identification. Facial recognition maybe? This should be interesting.
As a frequent traveler, this seems like a cost savings for airports at the expense of increasing my wait time in line behind inexperienced travelers.
DHS told us they are not using face recognition. They are using deep learning from ceiling mounted cameras that look straight down to re-identify people.
And hopefully it does cut down on wait time. This is part of their plan to automate screening and screen lower risk passengers less.
I would imagine that FR would be ill-suited for this kind of solution.
easier would be to assign ID tags to humans (easily identifiable as such by current analytics) with tags for the bags they are carrying/checking associated to the human tag.
i.e. human 1Q tag (H1Q) is associated with bag 1Q (B1Q)... H2R tied to B2R, etc
the problems they mention needing to be overcome are when these tag associations become tougher to apply directly.
It's kinda funny, as this project is very limited in its goal: only the passengers screening conveyance system, which COULD seem a rather simple application.
The luggages are quite separate one from the other, seen from the above with no occultation or lighting conditions issues, no crowd.
Of course, many luggages look the same so it's a bit more complex.
Also they are dealing with multiple cameras, though a luggage will not teleport from one lane to another, so there would be 3-4 cameras in line I would say.
What is funny is that there are tenders/manufacturers requesting or offering full public space reidentification programs, so this seems ridiculous in comparison.
However, I am yet to see an actually deployed program, has anyone encountered such a thing or is it still snake oil ?
Agree
Disagree
Informative
Unhelpful
Funny
Create New Topic
Login to read this IPVM report.
Why do I need to log in?
IPVM conducts reporting, tutorials and software funded by subscriber's payments enabling us to offer the most independent, accurate and in-depth information.
Comments (5)
Undisclosed Manufacturer #1
I wonder what analytic technology they will use to identify a human and search for a re-identification. Facial recognition maybe? This should be interesting.
As a frequent traveler, this seems like a cost savings for airports at the expense of increasing my wait time in line behind inexperienced travelers.
Create New Topic
Laurent ASSOULY
It's kinda funny, as this project is very limited in its goal: only the passengers screening conveyance system, which COULD seem a rather simple application.
The luggages are quite separate one from the other, seen from the above with no occultation or lighting conditions issues, no crowd.
Of course, many luggages look the same so it's a bit more complex.
Also they are dealing with multiple cameras, though a luggage will not teleport from one lane to another, so there would be 3-4 cameras in line I would say.
What is funny is that there are tenders/manufacturers requesting or offering full public space reidentification programs, so this seems ridiculous in comparison.
However, I am yet to see an actually deployed program, has anyone encountered such a thing or is it still snake oil ?
Create New Topic