analyzing customer behavior in retail space
@skalskip9250K followers
104,950 views
10 Comments
Comments from the original post on X (Twitter)

@skalskip92
dwell time tutorial, where I showed how to calculate how long detected objects stay within designated zones. link:
30 likes

@skalskip92
perspective conversion tutorial, where I showed how to calculate the real-world coordinates of a detected object. link:
19 likes

@giffmana
That’s only their location. There really interesting part is where they are looking at. You can implement this with this paper, works great, but does need some (easy) annotation:
53 likes

@skalskip92
how the hell do you know about all this papers? I thought you are doing VLMs now. Isn’t that full time job?
9 likes

@smtabatabaie
are there any good projects for detecting shoplifting recently?
3 likes

@skalskip92
I didn’t heated about anything :/
1 like

@hungrydonke
Wow! 🥕 Thanks! @skalskip92 It looks like my idea #4 got picked! This seems super useful for tracking areas of the supermarket that generate more or less interest and positioning products there based on their potential to catch customers' attention, and therefore budgeting campaigns with a scientific basis.
3 likes

@RedmondAI
It wouldn't be that hard to do pose estimation as well and then create a gaze-map which is arguably more important than just a location map.
2 likes

@skalskip92
Well I’m not sure which one is more important. Both 100% valuable to track customers behavior.

@deifosv
This in a casino must be interesting.
1 like