(PPS) Deformable Convolutional Networks

The basic idea of this paper is to give the convolution and pooling layers the ability to model different orientations and scales of objects in images. They do this by making the shape of the convolution filter learnable. For intuition, consider the…


(PPS) Cascade Residual Learning: A Two-stage Convolutional Neural Network for Stereo Matching

This paper tackles stereo matching which is the task of matching pixels from two images taken with two different cameras to deduce depth information. This is similar to…


Discrete Optimization: beyond REINFORCE

Gumbel Softmax (a.k.a. Concrete distribution)

While it might still be too early to start talking about the next A.I. winter, researchers are starting to see the limits of machine learning’s current set of…


(PPS) Dynamic Routing Between Capsules

Capsules is one of the recent works coming from Geoff Hinton which is why it’s so hyped up…


(PPS) DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

This paper deals with semantic segmentation: the task of labeling pixels in an image by their object classes. The goal of semantic…


(PPS) Efficient Deep Learning for Stereo Matching

Mini Distill
Mini Distill
This is a space where I write short summaries or extended tutorials of interesting papers I've read. The language is intended to be informal as to minimize the time between reading and understanding.
More information
Followers
235
Elsewhere