Deep Learning - Imaginea Labs

Shabda - A Neural Speech Denoiser

IntroductionIn speech processing, an important task encountered most often is the removal of noise from the given audio signal. Most of the time it happens that we are in situations where we want

GAN

A brief survey on GANs

Generative Adversarial Network (GAN) is one of the most interesting and popular class of generative networks in deep learning. GANs are finding applications in a variety of fields ranging from automated vehicles to

Reference

CycleGAN: How to do style transfer when there is no paired training data?

This paper presents an approach for learning to translate an image from a source domain X to a target domain Y in the absence of paired examples as shown below. Turning a horse

Reference

InfoGAN: A GAN that gives us control over the attributes of the generated images

In the Vanilla version of GAN, we do not have any control on the features of the images that are being generated. For example, with the Vanilla GAN trained over a collection of

Deep Learning

Visual classification of document images

Document image classification is not as well studied as natural image classification. We experimented with different neural network architectures on document image dataset. We discuss our preliminary results in this post.

Deep image prior

Deep Image Prior defies the idea that "deep learning only works in the context of massive datasets or models pretrained on such datasets". This paper showed that some deep neural networks could be

Deep Learning

Audio/Video receiver ports detection

Object detection process applied to Audio/Video receiver back panel images.

Deep Learning

Measuring feet using deep learning

Applying object identification techniques to a simple biometrics problem

Deep Learning

i-tagger - DL models for sequence tagging

Developing models to solve a problem for a data set at hand, requires lot oftrial and error methods. With current projects, we find a difficulty withsupporting different datasets and models in a modular way. i-taggerhelps with easing preprocessing, training and prediction.

Deep Learning

Deep Type

Given that we’re into typography, we would love to have a system where we can take a picture of some type and apply it to text of our own choice! We wanted to explore deep convolutional neural networks and use to build a new system to transfer style of typography.

Deep Learning

Typography detection

We talk about a filter we recently developed and deployed to reduce and flag incidences of non-typographic content on [Fontli], using a deep convolutional neural network based image classifier. We've had modest success and faced some intriguing situations and results along the way.