The Pascal VOC dataset is a standard benchmark for object class segmentation. Here we provide preprocessed data for use with the PyStruct structured prediction library. It consists of potentials derived from Philipp Krähenbühl's Textonboost implementation, superpixels computed with SLIC and superpixel connectivity graphs.
The LabelMe-12-50k data set contains 50.000 images of centered objects from 12 categories, which have been extracted from LabelMe.
For a description of the data set and for download go here.
The CUV Library is a C++ framework for easy use of Nvidia CUDA functions on matrices. It contains basic matrix operations and convolutions which can be performed both on the CPU or the GPU. The library also contains python binding so that all functionality can be used in python for fast prototyping.
The CUV Library can be downloaded at http://github.com/deeplearningais/CUV
We also provide a tutorial on an a
simple MLP that learns MNIST to get you started with CUV.
API Documentation is included, but can also be viewed online.
An open source implementation with NVIDIA CUDA™ that accelerates random forest training and prediction for image labeling by using the massive parallel computing power offered by GPUs.
Github link: https://github.com/deeplearningais/curfil