My results (yeah, a tough image) http://imgur.com/pbH52xW
From my experience with cats, "doormat" is actually pretty accurate. Damn things always dart right under my feet.
Deep Learning Image Classifier
21–30 of 33 posts
Re: Deep Learning Image Classifier
#22http://rekognition.com/demo/concept
Rekognition API has a similar API for all developers free.
It's reliable and very fast.
Checkout their demo page.
Re: Deep Learning Image Classifier
#23http://rekognition.com/demo/concept Rekognition API has a similar API for all developers free. It's reliable and very fast. Checkout their demo page.
I only tried one (hard) image, pizza-, sandwich- bloody mary https://imgur.com/30OgNdd. Rekognition seems to be working better than submission
Rekognition:
7.55% fruit; 0.92% dinner; 0.88% produce; 0.87% alcohol; 0.84% sliced
Toronto:
50% American lobster, Northern lobster; 12% plate; 7% crayfish, crawfish, crawdad; 7% Dungeness crab, Cancer magister; 4% king crab, Alaska crab; 4% butcher shop, meat market; 4% grocery store, grocery; 4% pomegranate
I find this interesting because I thought Hinton's group had state of the art tech. Who are these people and how do they do it?
Re: Deep Learning Image Classifier
#24What data was it trained on? Also can it tell you where in the image the identified object is?
This was pre-trained on ImageNet classes. You can find more information here: http://www.image-net.org
Re: Deep Learning Image Classifier
#25http://rekognition.com/demo/concept Rekognition API has a similar API for all developers free. It's reliable and very fast. Checkout their demo page.
I only tried one (hard) image, pizza-, sandwich- bloody mary https://imgur.com/30OgNdd . Rekognition seems to be working better than submission Rekognition: 7.55% fruit; 0.92% dinner; 0.88% produce; 0.87% alcohol; 0.84% sliced Toronto: 50% American lobster, Northern lobster; 12% plate; 7% crayfish, crawfish, crawdad; 7% Dungeness crab, Cancer magister; 4% king crab, Alaska crab; 4% butcher shop, meat market; 4% groce…
Lots of companies and teams are exploring deep neural network with all kinds of application. Rekognition API is the only one I found that provide open API service right now. You could train classifier using your own images. But you need to create an account and upload your images using their web application.
Re: Deep Learning Image Classifier
#26http://rekognition.com/demo/concept Rekognition API has a similar API for all developers free. It's reliable and very fast. Checkout their demo page.
I only tried one (hard) image, pizza-, sandwich- bloody mary https://imgur.com/30OgNdd . Rekognition seems to be working better than submission Rekognition: 7.55% fruit; 0.92% dinner; 0.88% produce; 0.87% alcohol; 0.84% sliced Toronto: 50% American lobster, Northern lobster; 12% plate; 7% crayfish, crawfish, crawdad; 7% Dungeness crab, Cancer magister; 4% king crab, Alaska crab; 4% butcher shop, meat market; 4% groce…
I think that when you're not expected to publish any papers to rationalize what you're doing, you're free to use any possible ugly hack to improve your results, (using a "kitchen sink" approach where you just combine the results of lots of unrelated techniques, extracting words from the URL, using the URL to actually fetch some related textual content on the website, etc). This gives private companies a competitive advantage over research institutions - their only purpose is to "make things work", not to introduce new techniques and have interesting insight about them.
Re: Deep Learning Image Classifier
#27President Obama is recognized either as ...
... a mountain-bike / all-terrain-bike (http://cdn2.spiegel.de/images/image-730849-galleryV9-vuuv.jp...)
... or a rugby ball (http://cdn2.spiegel.de/images/image-730849-breitwandaufmache...)
... or a bullet proof vest (http://cdn2.spiegel.de/images/image-730849-thumb-vuuv.jpg)
I guess the implementation leaves room for improvement :)
Re: Deep Learning Image Classifier
#28It seems strange that they would include in their set of example images, a picture of the most famous mausoleum in the world, without it being tagged with mausoleum or tomb or anything like that.
Re: Deep Learning Image Classifier
#29It seems strange that they would include in their set of example images, a picture of the most famous mausoleum in the world, without it being tagged with mausoleum or tomb or anything like that.
And it is tagged 99% mosque, while it isn't one. (the building on the left of it, not in the image, is).
Re: Deep Learning Image Classifier
#30There are several of these image classifiers now that someone should run an accuracy/speed/price comparison between
AlchemyAPI: http://www.alchemyapi.com/products/demo/alchemyvision/
UToronto
Rekognition: http://rekognition.com/demo/concept
Clarifai: http://www.clarifai.com/
I'm doing it myself, but I have a conflict of interest