There's a couple of indicators that at least for Arctic temperature, the numbers are headed south. First the weather plot from the drifting buoy that is connected with NOAA's North Pole Cam:
Artic temp graph 08-2009
© unknown

After some very brief excursions above freezing, it is now averaging below freezing. See the raw weather data here. The temperatures from the buoy have been hitting -2ยฐC regularly the past nineteen days.

Another indication is the north pole cam itself.
NOrth Pole Camera
© NOAA

Note that there are no melt pools or leads visible. The tilt is a bit puzzling, but as the temperature did get above freezing briefly, it may be a harbinger of things to come from this peer reviewed NASA paper. The buoy has drifted with the sea ice and is now near 84.1N, but started at 89.648N, so presumably, temperatures at the actual North Pole would be colder than what is being measured and seen now.

And finally the third temperature indicator is from the Danish Meteorological Institute.
Arctic temperature
© Danish Meteorological Institute

The blue 273.15K line is 0ยฐC or freezing. Of course seawater freezes a bit below that, typically at -2ยฐC according to the US Navy.

The DMI plotting model, seen here, comprises the "daily mean temperature and climate north of the 80th northern parallel, as a function of the day of year." Note that this year, the temperature rise got off to a late start. However the drop below freezing appears to be headed along the normal bell curve.

While sea ice extent has not seen the depths of 2007 or 2008 numbers yet, it did take a small hit to extent numbers yesterday. Given the temperature situation, it may begin to take a turn in the next two weeks. Traditionally, we have about 20 days of melt season left from this point.
Artic Sea Ice Extent
© IARC-JAXA

However, as Jeff Id noted in his sea ice video, the winds of the polar region seem to be more of a determining factor in extent than temperature, as much of the ice is being driven southward. Our NOAA North Pole Cam and its drift from 89.648N to 84.1N is an example of this movement.

The date of the turn is always interesting to speculate. I'm going to take a SWAG and suggest Sept 9th as the day Arctic sea ice extent bottoms out.