79414939
submission
subh_arya writes:
Researchers from NYU, UToronto and MIT have come up with a technique that captures human learning abilities for a large class of simple visual concepts to recognize handwritten characters from World's Alphabet. Their computational model represents concepts as simple programs that best explain observed examples under a Bayesian criterion. Unlike recent deep learning approaches that require thousands of examples to train an efficient model, their model can achieve human-level performance with only one example. Additionally, the authors present several “visual Turing tests” probing the model’s creative generalization abilities, which in many cases are indistinguishable from human behavior. A science magazine article is here.
77560531
submission
subh_arya writes:
Automatic crowd counting has been an extremely challenging computer vision problem. However, researchers from UCF, seem to have found a reasonably accurate solution using sophisticated probabilistic models. Although there has been several previous efforts in this direction, this is the first time the technology has been put to use on a realistic scenario where around 550,000 protesters participated for Catalunyan Independence. A freely available technical paper published in IEEE Trans. on Pattern Analysis and Machine Intelligence, 2015 is available here.
75670579
submission
subh_arya writes:
Researchers from Microsoft Research unveil the first technology to perform 3D surface reconstruction from ordinary smartphone cameras. Their computational framework creates a connected 3D surface model by continuously registering RGB input to an incrementally built 3D model. Although the reconstruction results look promising, Microsoft does not claim to release an app anytime sooner.
