After 10 weeks of climbing steep hills in Hampstead and Highgate, my confidence soared that I could better my time from last year. How complacent I was. In the end, I didn't even make it to Nice.

The weather was predicted to be very hot with a slight headwind from the South. There was also a massive cold front gathering momentum over the Atlantic. It was expected to reach Nice within the next few days, so the pressure was on to keep up the pace. (I would eventually watch it rolling over France from the safety of the TGV five days later). Though the headwind wasn't particularly strong the first two days, it blew consistently and strengthened in the Rhone valley. It slowed me down quite significantly, perhaps by an hour each day. As I woke up in Baix near Montelimar, 48h after leaving Paris, the idea thus grew on me that I ought to take it more leisurely and end my journey in Avignon, a town I had always wanted to visit anyway! And so I did, instead of veering east after Orange towards Aix-en-Provence and into a third night, I continued south on the N7 and rolled into Avignon before the day was over.

Here is the route on Google Maps. The first two stages are the same as before with 360km from Paris to Digoin the first day, and 260km from Digoin to Baix the second day.

I didn't vary the equipment from last year, except for two new Schwalbe Durano tyres that kept their promise and afforded me a puncture-free ride. As I had been doing this trip quite a few times, I thought I could start riding it from memory (instead of carrying dozens of Google Map print-outs covering critical sections of the journey). How wrong I was. In pretty much every bigger city, I took at least one wrong turn and ended up losing hours to my own complacency. Perhaps it's time, at last, to get a GPS.

We got a paper on contextual object classification into Australia's annual conference on digital image computing - DICTA 2010. It's in Sydney this year, and I wished I could attend!

My personal Paris-Nice in 3 days cycling challenge took place end of June. It has been an amazing three days on the bike and I look forward to repeating it in 2011. Every year I try to optimise the route further and it seemed to have paid off this year. Latest route advice therefore:

• Between Nevers (after 250km) and Digoin (after 350km) stay south of the Loire as it's not only very flat but also considerably more quiet than on the departmental road a few kilometers to the north. As the road runs close to the bottom of the wet Loire valley, the landscape becomes covered in dense fog soon after temperatures start to drop. The atmosphere is really rather magical as the mist, softly lit from above by the moon, envelops everything bar the canopies of the tallest trees, which are the only visible structures that rise above this dreamy nightscape.
• Once in the Rhone valley at Andance, stay west of the river until almost 40km before Orange, and thus avoid the big cities Montelimar and Valence. The road is again much more quiet and, after dawn, a lot more scenic.
• Once at the Mediterranean in Frejus, take the beautiful, if steep, mountain road through the Foret d'Esterel instead of following the gruelling ups and downs along the coast. It is one 25km climb up and then downhill all the way to Cannes. A magnificent last stretch before the flat finale to Nice.

It was the first time I rode without my shimano flightdeck or any other speedometer, something I had previously considered psychologically insufferable. But I loved it. It was only afterwards that I figured out how much I had cycled each day: a good start with 350km on the first day, just over 250km on the second day, another 310km the third day, and 50km to the finish line in the early morning hours of the fourth day. I equalled the time from my 72h maiden ride in 2002, much of which I had spent getting lost or recovering in cafes, something I tried to minimise this time. It is safe to conclude that I was much fitter back then. Some pictures of the Paris-Nice trip.

A video from our Spring trip to Morocco, produced by one of the other pilots.

and a few paragliding pictures from the morocco 2010 trip.

This conference paper is a simple extension of the 1-bit quantisation of wavelet coefficients proposed by Jacobs, Finkelstein and Salesin in the mid-90s. We show that a quantiser with four partition regions improves retrieval results and prove a few theorems that help to find optimal partition sizes.

We developed an unsupervised version of Grabcut in which the initialisation step is automated, and propose a new stopping criterion based on the KL divergence between background and foreground models.