Convolutional Neural Networks レシェック　リビツキ leszek@abeja.asia

xkcd 24.09.2014

Is there a bird in the picture?

ImageNet Large Scale Visual Recognition Challenge 1. Image classification (2010-2014): What’s in the picture? 2. Single-object localization (2011-2014): Mark one object of each category 3. Object detection (2013-2014): Mark all objects of each category

abacus, abaya, academic gown, accordion, acorn, acorn squash, acoustic guitar, admiral, affenpinscher, Afghan hound, African chameleon, African crocodile,African elephant, African grey, African hunting dog, agama, agaric, aircraft carrier, Airedale, airliner, airship, albatross, alligator lizard, alp, altar, ambulance, American alligator, American black bear, American chameleon, American coot, American egret, American lobster, American Staffordshire terrier, amphibian, analog clock, anemone fish, Angora, ant, apiary, Appenzeller, apron, Arabian camel, Arctic fox, armadillo, artichoke, ashcan, assault rifle, Australian terrier, axolotl, baboon, backpack, badger, bagel, bakery, balance beam, bald eagle, balloon, ballplayer, ballpoint, banana, Band Aid, banded gecko, banjo, bannister, barbell, barber chair, barbershop, barn, barn spider, barometer, barracouta, barrel, barrow, baseball, basenji, basketball, basset, bassinet, bassoon, bath towel, bathing cap, bathtub, beach wagon, beacon, beagle, beaker, bearskin, beaver, Bedlington terrier, bee, bee eater, beer bottle, beer glass, bell cote, bell pepper, Bernese mountain dog, bib, bicycle-built-for-two, bighorn, bikini, binder, binoculars, birdhouse, bison, bittern, black and gold garden spider, black grouse, black stork, black swan, black widow, black-and-tan coonhound, black-footed ferret, Blenheim spaniel, bloodhound, bluetick, boa constrictor, boathouse, bobsled, bolete, bolo tie, bonnet, book jacket, bookcase, bookshop, Border collie, Border terrier, borzoi, Boston bull, bottlecap, Bouvier des Flandres, bow, bow tie, box turtle, boxer, Brabancon griffon, brain coral, brambling, brass, brassiere, breakwater, breastplate, briard, Brittany spaniel, broccoli, broom, brown bear, bubble, bucket, buckeye, buckle, bulbul, bull mastiff, bullet train, bulletproof vest, bullfrog, burrito, bustard, butcher shop, butternut squash, cab, cabbage butterfly, cairn, caldron, can opener, candle, cannon, canoe, capuchin, car mirror, car wheel, carbonara, Cardigan, cardigan, cardoon, carousel, carpenter’s kit, car- ton, cash machine, cassette, cassette player, castle, catamaran, cauliflower, CD player, cello, cellular telephone, centipede, chain, chain mail, chain saw, chain- link fence, chambered nautilus, cheeseburger, cheetah, Chesapeake Bay retriever, chest, chickadee, chiffonier, Chihuahua, chime, chimpanzee, china cabinet, chiton, chocolate sauce, chow, Christmas stocking, church, cicada, cinema, cleaver, cliff, cliff dwelling, cloak, clog, clumber, cock, cocker spaniel, cockroach, cocktail shaker, coffee mug, coffeepot, coho, coil, collie, colobus, combination lock, comic book, common iguana, common newt, computer keyboard, conch, confectionery, consomme, container ship, convertible, coral fungus, coral reef, corkscrew, corn, cornet, coucal, cougar, cowboy boot, cowboy hat, coyote, cradle, crane, crane, crash helmet, crate, crayfish, crib, cricket, Crock Pot, croquet ball, crossword puzzle, crutch, cucumber, cuirass, cup, curly-coated retriever, custard apple,...

No content

No content

Better than human? In the object detection with external data track, the winning team was GoogLeNet (...). It is truly remarkable that the same team was able to win at both image classification and object detection, indicating that their methods are able to not only classify the image based on scene information but also accurately localize multiple object instances. [...] We found that humans are noticeably worse at fine-grained recognition (e.g. dogs, monkeys, snakes, birds), even when they are in clear view. To understand the difficulty, consider that there are more than 120 species of dogs in the dataset. We estimate that 28 (37%) of the human errors fall into this category, while only 7 (7%) of GoogLeNet errors do. Olga Russakovsky*, Jia Deng*, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, Alexander C. Berg and Li Fei-Fei. (* = equal contribution) ImageNet Large Scale Visual Recognition Challenge. IJCV, 2015. http://arxiv.org/abs/1409. 0575

How did they do it?

Visual Cortex Source: A Model of V4 Shape Selectivity and Invariance Charles Cadieu, Minjoon Kouh, Anitha Pasupathy, Charles E. Connor, Maximilian Riesenhuber, Tomaso Poggio Journal of Neurophysiology Published 1 September 2007 Vol. 98 no. 3, 1733-1750 DOI: 10.1152/jn.01265.2006 http://jn.physiology. org/content/98/3/1733

Blakemore’s cats Blakemore, Colin, and Grahame F. Cooper. “Development of the brain depends on the visual environment.” (1970): 477-478.

Convolution filters https://docs.gimp.org/en/plug-in-convmatrix.html

Convolution layer Source: Neural Networks and Deep Learning By Michael Nielsen / Jan 2016 http://neuralnetworksanddeeplearning.com/ Chapter 6: Deep Learning

Subsampling/ Max pooling Convolutional Neural Networks (CNNs / ConvNets) http://cs231n.github. io/convolutional-networks/#pool

Park or bird network Flickr PARK or BIRD http://parkorbird.flickr.com/ http://code.flickr.net/2014/10/20/introducing-flickr-park-or-bird/

Deep Convolutional Neural Network

Supervised learning

What else is possible?

Deep dream

Coloring black and white photos and movies

Artistic Style Transfer

Counting How many people in the shop?

Demographic What gender and age are they?

Behavior Where do they spend time most?

...and more!

Are you interested in: ● Convolutional Neural Networks? ● Deep Learning? ● Big Data? ● Internet of Things? ● the FUTURE? recruit@abeja.asia