This idea cunningly uses the ubiquity and pseudo- random distribution of street litter to make deductions about the sources of that litter and maybe find the very fast food restaurant whose rubbish you've been analysing.

You would need a video camera, a computer system and a display and to use it you would point the camera at the pavement as you walk around an urban area. You would be looking ahead, holding the display in your field of vision.

The output of the camera would processed in real time to order to identify the type and brands of street litter, recording their positions and noting additional information like clustering. There would also be a data storage, black box feature for longer-term off-line analysis.

For this real-time feed the display would be used to recommend a pedestrian search strategy to find the place that a branded bit of litter comes from. As you wander, using the display's advice (but watching out for traffic, little old ladies and the like at the same time) your video input is constantly collecting more information to refine it's search pattern, trying to detect if you are getting closer or further away.

Initially it would be a way to test whether an algorithm like this is possible but in the end it could be used as an urban tracking device to find business or people who are prone to litter or encourage littering through excessive packaging.

I feel it in my bones (and my properly disposed of burger wrapper) that this is possible. Of course clean streets, or a complete lack of pedestrians, would render this device useless.
-- Aristotle, May 04 2009

The Garbage Project http://traumwerk.st...u:3455/symmetry/174
"The largest category occupying landfill space was paper." [normzone, May 04 2009]

I agree with both clauses of your last paragraph; I think this is possible, but it needs a great deal more littering to be effective. Admittedly I do not exhaustively look at and classify litter, but I don't see a lot of it generally, and what there is mostly seems to lack strong connection to nearby businesses .... of course, your mileage may vary. [+] anyway.
-- batou, May 04 2009


(misplaced apostrophe in the subtitle)

If this were widely used might it not encourage fast-food outlets to generate more litter as a means of attracting customers? - hang on, they probably do that anyway... [+]
-- hippo, May 04 2009


(+)

I'll wager somewhere there's a student who has done an analysis of all the trash picked up in a given zone - I wonder if there are any documented findings.

So this could be a debris-based mixture of Google Earth and Mapquest.

Both wind and human traffic patterns would bias the data, but I can just imagine:

"Click here to plan your trip to include an In-And-Out-Burger place probably right about here".

In Samuel R. Delany's "Stars in My Pockets Like Grains of Sand" the professionals that swept the streets, hauled the garbage and studied their collections for patterns were called "tracers" and the position was held in some esteem.
-- normzone, May 04 2009


Analysis of rubbish is always good in my books. An idea too minor for the Halfbakery is my petty idea to tax companies by the weight of their branded rubbish that their customers discard onto the streets.

So this is more of an social, environmental and computational algorithm that we can use in what seems to be (for me at least, perhaps I'm just too grumpy) an age of rubbish. Understand the object rich environment now, track down the chain smoking murder suspect later ...

The gain might be a better understanding of rubbish and another gain for object recognition systems. Some of those packaging items seems to be well documented due to the various trademarks and copyrights that they declare in between the patterns.
-- Aristotle, May 04 2009


Most archaelogy is based on the analysis of trash.
-- DrBob, May 05 2009



random, halfbakery