Profile cover photo
Profile photo
Ludwig Schmidt-Hackenberg
106 followers -
German Engineer interested in computer vision, content based image retrieval and adventure.
German Engineer interested in computer vision, content based image retrieval and adventure.

106 followers
About
Posts

Post has attachment
Do you want to do Image Retrieval on Millions of images. Join EyeEm as a Computer Vision Engineer or Intern. http://www.eyeem.com/careers 

Post has attachment
In love with photography & computer vision? Join +Appu Shaji and me EyeEm as Product Manager for our new image search. http://www.eyeem.com/careers 
Add a comment...

Post has shared content
Nice summery of the ImageNet 2013 results from LeCun.
The results of the 2013 ImageNet Large Scale Visual Recognition Challenge are out!

The NYU teams did quite well (yes, there are several NYU teams).

The competition had three components
- Classification: ImageNet dataset with 1000 categories
- Classification+Localization: same, but one must provide a bounding box
- Detection: 200 categories, possibly several objects per image

On classification, Matt Zeiler's "Clarifai" system won with less than 12% error (top 5), followed by NUS, Andrew Howard, the Zeiler-Fergus team (NYU), the OverFeat team (NYU), and the University of Amsterdam-Euvision team. Other teams were above 15% error.

Matt Zeiler graduated a few weeks ago from his PhD with +Rob Fergus at NYU. Matt has been playing with ImageNet classification for about 1 year, and had a chance to fine-tune his system. The ZF system is what he built with Rob, while the Clarifai system is what he built since he graduated. 

The OverFeat team (composed +Pierre Sermanet,  David Eigen, Michael Mathieu, +Xiang Zhang, +Rob Fergus, and myself) has had less time to tune it classification entry, and concentrated on the classification+localization competition. We won that one handily, but there was only one other entry (but a mighty one: VGG/Oxford).

Our OverFeat entries are fairly standard convolutional networks (surprise!) with a few training tricks. Our best entry (14.1% error) is a committee of 7 convnets, while our second entry (15.6%) is a single convnet. Unlike many of the other teams, we use our own GPU implementation interfaced to the Torch7 package (http://torch.ch) and we do not use Alex Krizhevsky's GPU code.

Our OverFeat team also participated in the detection challenge. We did pretty well, but our numbers were still increasing quite rapidly by the time of the deadline, and we think we can do much better. The detection challenge was won by U of Amsterdam-Euvision, with NEC and OverFeat closely behind. The other teams are quite some ways behind. Our system was pre-trained with the ImageNet-1K before being fine-tuned on the 200-category detection dataset (which is why it appears on a dark grey background in the table of results).

#ImageNet   #convnet   #Torch7   #computervision   #machinelearning   #deeplearning  
Add a comment...

Post has attachment
Blogpost: The end of Everpix, a sad week for photographers and machine learning 
Add a comment...

Post has shared content
Videos of IPAM Summer School on Deep Learning and Feature Learning

The videos of the lectures of the 2012 IPAM Graduate Summer School on Deep Learning and Feature Learning are finally available.
(It's taken a bit of time because of technical issues).

This constitute the most complete set of talks on deep learning, convolutional nets, unsupervised feature learning, sparse coding optimization for deep learning, theoretical results, practical issues, connections with neuroscience.....

There are talks by +Geoffrey Hinton, +Yann LeCun+Yoshua Bengio, +Andrew Ng, +Rob Fergus, +Alan Yuille, +Graham Taylor, Stephen Wright, Arthur Szlam, Stephane Mallat, +Kai Yu, +jorge nocedal, +Jason Morton, Stan Osher, +Russ Salakhutdinov, +Marc'Aurelio Ranzato, +Jason Weston, +Bruno Olshausen, +Thomas Serre, +Roland Memisevic, +Iain Murray, and +Nando de Freitas.

A pretty impressive collection of speakers, don't you think?           
Add a comment...

Post has shared content
You can use huge gifs as your g+ cover image.
New cover photo. Click profile to see. Gifalicious. 
Add a comment...

Post has shared content
In case you haven't read it already, +Geoffrey Hinton and two of his PhD students, +Ilya Sutskever and +Alex Krizhevsky, are joining Google.
Last summer, I spent several months working with Google’s Knowledge team in Mountain View, working with Jeff Dean and an incredible group of scientists and engineers who have a real shot at making spectacular progress in machine learning. Together with two of my recent graduate students, Ilya Sutskever and Alex Krizhevsky (who won the 2012 ImageNet competition), I am betting on Google’s team to be the epicenter of future breakthroughs. That means we’ll soon be joining Google to work with some of the smartest engineering minds to tackle some of the biggest challenges in computer science. I’ll remain part-time at the University of Toronto, where I still have a lot of excellent graduate students, but at Google I will get to see what we can do with very large-scale computation.
Add a comment...

Post has attachment
Still confused by SVMs & these Kernel thingies? Listen to Pedro Domingo's Coursera lecture on machine learning. He gives the best introduction to SVM I've found so far.
Add a comment...

Post has attachment
SMBC explains what will happen when we finally have the technology to upload our brains to computers.
Add a comment...

Post has shared content
Nice!
#ifIhadglass, we would make it possible to see through time, by
automatically aligning (using algorithms we developed) the
present-day landmarks (e.g. entire historical center of Paris) with
archival footage and historical paintings, producing a large-scale
immersive historical experience (with +Josef Sivic and +Ivan Laptev )
Photo
Add a comment...
Wait while more posts are being loaded