Starting this week Deep AI is providing automated AWS AMI builds of the open-source deep learning server Deepdetect (http://www.deepdetect.com/). The AMI is bundling Deepdetect with a set of state-of-the art deep learning models ready to be used for common tasks such as e.g. gender detection. The AMI also comprises a directly usable web frontend to display results from deep learning models.
The AMI can be instantiated and used by anyone with an Amazon AWS (https://aws.amazon.com/) account. You only need the following:
- some curiosity and excitement about deep learning
- a computer running Windows, Mac OSX or Linux with internet connectivity
- a credit card and willingness to spend at least 1 EUR on AWS computing ressources (each 1 EUR is buying you computational ressources for more than 1 hour from Amazon for your experiments; the AMI itself as well as Deepdetect is free of charge)
In fact it is really easy to set up the AWS account, instantiate the AMI and run a deep learning based prediction service!
The beauty about Deepdetect when compared with some commercial deep learning APIs is that it is – being open source and integrating Berkeley’s Caffe (http://caffe.berkeleyvision.org/) – fully configurable and can be used for any state-of-the art deep learning related vision task. You might train your own model for your very specific task and directly publish your results via REST API to everyone interested.
So what do you need to do? Detailed descriptions about setting up the AMI and starting and running deepdetect on AWS are provided by one of our colleagues at Deep AI using the github user name revilokeb: https://github.com/beniz/deepdetect/issues/5.
If you have got any questions feel free to contact us!