Nicolas Pinto
Establishing Good Benchmarks and Baselines in Artificial and Biological Vision
Abstract
Progress in understanding the brain mechanisms underlying vision requires the construction of computational models that not only emulate the brain's anatomy and physiology, but ultimately match its performance on visual tasks. In recent years, ostensibly "natural" images have become popular in the study of vision, and have been used to show apparently impressive progress in building such models. In this talk, we will demonstrate that tests based on uncontrolled natural images can be seriously misleading, potentially hindering progress and guiding the community in the wrong directions. Instead, we re-examine what it means for images to be natural and argue for a renewed focus on the core problem of object recognition -- real-world image variation.
Short Bio
Nicolas Pinto is currently Chief Scientist and Chief Technology Officer of two Silicon Valley stealth startups, focusing on the research and development of human-level brain-inspired perception technologies and their real-time applications on low-power embedded devices. He holds two M.S. in Computer Science and Engineering from France (UTBM/ENSISA, 2007), and a Ph.D. in Neuroscience from the USA (MIT, 2010) supported by NSF, DARPA, NVIDIA, Google, Amazon and Microsoft. Previously he was a graduate-level Lecturer in Computer Science at Harvard SEAS/DCE teaching Massively Parallel Computing, and a Research Scientist in Prof. Jim DiCarlo's Lab at MIT and Prof. David Cox's Lab at Harvard developing large-scale computational models of the visual cortex.
Nicolas was the IEEE CVPR 2011 publication chair, served on the program committee at ACM InPar 2012, IEEE CIGPU 2012, NIPS ”Big Learning” 2011, GPGPU 2011-2012, IEEE ECCV/CVGPU 2010, PASCO 2010, and reviewed for IEEE PAMI, NIPS, IEEE Transactions on Image Processing, Elsevier Vision Research, IEEE IJCNN, IEEE ICDL, etc.
Relevant papers
Pinto N, Cox DD, DiCarlo JJ - Why is Real-World Visual Object Recognition Hard? (PLoS 2008)
http://goo.gl/6Kizr
Pinto N, Dicarlo JJ, Cox DD - Establishing Good Benchmarks and Baselines for Face Recognition (ECCV 2008)
http://goo.gl/RqjE2
Pinto N, Doukhan D, DiCarlo JJ, Cox DD - A High-Throughput Screening Approach to Discovering Good Forms of Biologically-Inspired Visual Representation (PLoS 2009)
http://goo.gl/dcnN8
Pinto N, Dicarlo JJ, Cox DD - How far can you get with a modern face recognition test set using only simple features? (CVPR 2009)
http://goo.gl/MmYV7
Pinto N, Majaj NJ, Barhomi Y, Solomon EA, Cox DD, DiCarlo JJ - Human versus machine: comparing visual object recognition systems on a level playing field. (COSYNE 2010)
http://goo.gl/Yk5uL
Pinto N, Cox DD - An Evaluation of the Invariance Properties of a Biologically-Inspired System for Unconstrained Face Recognition (BIONETICS 2010)
http://goo.gl/K7qfw
Pinto N, Barhomi Y, Cox DD, DiCarlo JJ - Comparing State-of-the-Art Visual Features on Invariant Object Recognition Tasks (WACV 2011)
http://goo.gl/EvvV0
Pinto N - Forward Engineering Object Recognition: A Scalable Approach (MIT PhD Thesis 2011)
http://goo.gl/UVv09
Cadieu CF, Hong H, Yamins D, Pinto N, Majaj NJ, DiCarlo JJ - The Neural Representation Benchmark and its Evaluation on Brain and Machine (ICLR 2013)
http://goo.gl/ORC3Y