Depth Estimation

We percieve world in 3D, machines do the same in pixel values, inorder to bridge this gap Depth Estimation is formulated and still is an active area of research.

Adversarial Training

State-of-the-art neural networks show execllent generalization, some times even across domains, but are susceptible to structed adversarial perturbations let's review works on them.

Domain Adaptation

Deep Neural Network show excellent performance when the train and test data distribution are close but fails to generalize across the domains unlike humans, let's review papers which address this....

Style Transfer

Imagine if Picaso or Da vinci is alive today and you can get yourself painted by them, well style transfer works can just do that, let's review some of them....

Video Object Segmentation

Segmenting a desired object from a scene is a basic computer vision task, but doing it on a video requires temporal consistancy, and can be done better using the information...

Super Resolution

Have you ever wondered if you can increase your old photographs resolution with the new technology, let's review how to do that.

GAN

Generative Adversarial Networks contains one generator network which generates the data(either from random noise or some input) and a discriminator network which differentiates the predictions from the ground truth, let's...

Attention

Humans have a very good algorithm to prune the vast amount of data that comes in to the required data essential for making decisions, let's review the works that emulate...

Sketch

Sketch images capture class information and the object information using very few strokes, humans very easily understand sketches unlike neural nets, let's review the works which make deep nets understand...

Deep Learning Courses and People

Few Resources for starting with deep learning, legenedary people in the area and few professors...

1 minute read