After a fair bit of graft we’ve finished our first product using StrongSteam – a Latin Botanical Garden label matcher (AKA “OpenPlants”) which runs at Kew Gardens, Wakehurst Place and other botanical gardens in Europe that use the usual black rectangular labels. If you’re not sure what I’m talking about then these 30 second demo videos should make it clear:
Update – I totally should have added that we built this app in partnership with Kasabi using their DBPedia dataset.
As you can see you photograph the plant label, we use optical character recognition to read what’s on the sign and then we bring back relevant information from wikipedia about the Family and Genus including pictures and links to other resources. We’ll launch this as a free app in a few weeks.
Seeing as StrongSteam is a cloud based API it makes sense to show it being used from another platform. Here’s a screencast showing a webcam on a Linux laptop taking a photograph of a printed plant label using our Python API which is uploaded and recognised, with the results being shown in the local web browser:
We’ll launch the alpha OCR API for developers in April. Add your email to the email list on our homepage to get an announce. Once the iPhone app is available we’ll also announce it here.
I’m very chuffed to say that our 2nd AI/Data/API meetup will take place next Wednesday 4th April 7pm at the Santiago MakerSpace at Avenida Italia 850 (map). The HackerSpace is a 15 minute walk south/east from the CMI office, just off of Bilbao.
UPDATE we’re meeting at 7pm (not 7.30pm) and we have an extra speaker – Skype co-founder Ahti Heinla will tell the “Real story behind the Skype success story” in an extra 30 minute slot. Here’s the Meetup announce with full details. Remember to Attend if you’re coming and do UnAttend if you realise you can’t make it later (the numbers are limited).
The event is free (but bring cash for the specialist beer – see end of this post). Our four speakers will talk for 15 minutes each:
Tim of BackYardBrains – “Curious about how your brain works? With the help of our friends the humble Amazonian cockroach, we will teach you the electrical properties of neurons.” – will include demos and hardware you can buy (angellist)
Ashley of PaperHater will give a live demo of their receipt/paper scanning and OCR application and talk about how they created a spinoff
Javier Gramajo of SQMOS will give a live demo of their augmented reality Android app (angellist)
StrongSteam and BackYardBrains are Round 2, PaperHater is in Round 3 and SQMOS are staying in Santiago after Round 1. Stick around after to learn about the hackerspace and drink Ignacio’s specialist beers.
The Maker/Hacker Space is the first in Santiago, here’s some news about their launch and they have a circuit building workshop on 6-8th April. In this space you will find:
Robots with legs and robots that fly
8 bit computers
Dirty work space (CNC lathe and other construction equipment)
Lots of creative equipment that you can use to build electronic, mechanical, art and music things if you choose to become a member
Ignacio Correa of ClubCervezas will bring along specialist Chilean beers (to buy) – he took us on a beer tasting night in Bario Brasil a couple of weeks back and introduced us to a whole range of lovely local beers (300 microbreweries!) that we hadn’t tasted before. Bring some money if you’d like to try unusual and hard-to-find Chilean beers (my pockets will be stuffed with cash).
It has been a pretty nutty couple of weeks. PyCon a week back was ace, we signed up some clients and partners for StrongSteam and got offered investment. David Kim was good enough to interview me so I got to demo our OCR for text recognition and image recognition APIs via some mobile demos – check out the second video on David’s Enthought post.
Check out the video below at 0:19:00 to 0:32:00 to see me pitch and then at 1:05:00 to 1:07:30 to see the three judges decide that StrongSteam was ‘best bet for investment’. Being judged was fun. Focusing on giving our users what they need from our API is more our focus for now.
A few days before I was submitting the second month of reimbursement paperwork for our StartupChile placement. Emily has written a long piece on this already.
Below you can see my pile of paperwork – for each transaction (few big purchases, some contractors, some travel) I have a full audit trail that starts at the receipt and ends, via banks and credit cards, to a bank account in my name, with proof that I own that bank account. For contractors I include a full contract too. This proof is required, this is the ‘price’ of giving up 0% equity under a government scheme. It took 8 hours including my meeting with my account executive. They haven’t reimbursed this round yet, assuming they don’t reject anything (which is far from guaranteed) then this only costs 8 hours (last month cost 2 days). If they reject stuff then maybe I’ll invest a total of 10-16 hours.
Something that’s painfully obvious from yesterday’s pitching and today’s BizCamp is that pretty much all of us here lack t-shirts with our name, logo & strap-line. I could really have done with t-shirts at PyCon, I pitched to 100+ of the 2,300 delegates but got on stage in front of them all once – if someone had seem our name and noticed ‘AI’ or ‘computer vision’ then I bet they’d have come over for a chat. Lesson learned.
I’m also going to give a shout out back to Moo in the UK for their cool little business cards. So many people here don’t have any cards yet, this is such a mistake. Everyone needs cards, I’ve used Moo for years, I’d vote you go via them and get the mini cards and a plastic case (they’re robust, mine is >2 years old and is still fine).
Finally – Vivek Wadhwa kicked a bunch of us up the arse two nights ago and again last night talking about self-mentorship (given that there is no formal mentorship out here). I’m going to be organising a group who want to self mentor such that we can meet regularly (maybe every week), set goals, be held accountable and basically focus on getting ready for demo day in 2 month’s time. It’ll be an interesting experiment.
For now this is nearly the end of a crazy 2 months. Tonight I’m going to get a take-out Chinese and settle in front of a movie.
UPDATE I’m thinking of writing an updated guide (possibly as a self-published book), if you’re interested in hearing about it please join the High Performance Python Mailing List (I’ve only got a list right now). I’ll make an announce once I know more.
Profiling with cProfile and line_profiler
Profile visualisations with runsnake
PyPy for quick wins
Cython for C-level speed
ShedSkin for ‘quick wins’ on the right problems
Cython+numpy for multi-core (300* on this Mandelbrot problem) speed-ups
Multiprocessing for multi-core support
ParallelPython for multi-machine support
Numexpr for faster numpy math
The other topics in this high performance track (a part of the tutorial track) are:
After EuroPython I wrote up my talk with additional material as a 55 page book, I was hoping to update the book this year but things are moving so fast with our new StrongSteam AI/vision startup (presented at StartupRow at PyCon) that I can’t really justify the time right now. I’ll just link to the High Performance Python book from last year, the timings are out of date (but they’re correct in the slides below) and the src is updated a bit, but the method and discussion is still correct.
PyCon 2012 is just coming to a close. There were over 2,200 people here and too many talks to choose between. It was a bloody fine conference. Meeting so many of the Names of the Python world was rather grand, teaching High Performance Computing and getting pats on the back for the creation of ShowMeDo was also rather nice.
We won a booth on StartupRow on the Friday in the expo hall, I adorned our stand with some posters and props for our mobile phone demos. With StrongSteam we’re working to give a pair of eyes to mobile phones so phones can ‘see’ the world as a human does. What could you do with an API that let’s you build your own Google Goggles?
Kyran and Balthazar had put together some cool demos – OCR on photos of labels to read text and open relevant wikipedia entries and also artwork recognition for Shardcore‘s art. We won a few offers of angel investment, had an acquisition offer, got some users and found some collaborators. Not bad for the second day at the conference.
One criticism I have is that StartupRow wasn’t advertisied. We were given small booths at the back of the hall behind the big shiny stands so it looked a bit like we were the poor cousins to the ‘proper’ companies. A banner or other announcement priming folk to the idea that we were early stage would have been handy.
Today the expo hall was cleared for the poster session. This was huge, I was very happy to see a wide selection of science and HPC projects along with a handful of companies.
Now I need to sleep before the 4am wakeup for the return flight to Santiago. Then…finishing off our first StrongSteam client and moving towards inviting users into the API.